diff --git a/backend/app/routers/websites.py b/backend/app/routers/websites.py
index de6d954..281bf0d 100644
--- a/backend/app/routers/websites.py
+++ b/backend/app/routers/websites.py
@@ -509,12 +509,15 @@ def reorder_account_priorities(
     failed_count = sum(1 for item in results if item.get("status") == "failed")
     success_count = sum(1 for item in results if item.get("status") == "success")
     skipped_count = sum(1 for item in results if item.get("status") == "skipped")
+    stale_count = sum(1 for item in results if item.get("status") == "stale")
 
     parts = []
     if success_count:
         parts.append(f"更新 {success_count}")
     if failed_count:
         parts.append(f"失败 {failed_count}")
+    if stale_count:
+        parts.append(f"清理失效账号 {stale_count}")
     if skipped_count:
         parts.append(f"跳过 {skipped_count}")
     message = "、".join(parts) + f" / 共 {len(results)} 个" if parts else "没有需要重排的账号"
diff --git a/backend/app/services/website_client.py b/backend/app/services/website_client.py
index b99135d..1dd2c01 100644
--- a/backend/app/services/website_client.py
+++ b/backend/app/services/website_client.py
@@ -225,7 +225,12 @@ class Sub2ApiWebsiteClient:
 
     def update_account(self, account_id: str, body: dict[str, Any], endpoint: str = "/accounts") -> dict[str, Any]:
         """更新远端账号（仅传入需要变更的字段）。"""
-        resp = self._request("PUT", f"{endpoint}/{account_id}", body)
+        try:
+            resp = self._request("PUT", f"{endpoint}/{account_id}", body)
+        except WebsiteError as exc:
+            if isinstance(exc.__cause__, httpx.HTTPStatusError) and exc.__cause__.response.status_code == 404:
+                raise WebsiteError(f"目标账号 {account_id} 不存在或已被删除") from exc.__cause__
+            raise
         data = _unwrap_data(resp)
         return data if isinstance(data, dict) else {"value": data}
 
diff --git a/backend/app/services/website_sync.py b/backend/app/services/website_sync.py
index 936df62..fa61642 100644
--- a/backend/app/services/website_sync.py
+++ b/backend/app/services/website_sync.py
@@ -472,74 +472,13 @@ def sync_account_priorities_for_upstream(
             # 不写日志、不发通知：网站不可用时账号无法更新，沉默跳过
             continue
 
-        # 查询该网站所有已导入 Key（跨上游），用于倍率查询
-        all_website_keys = (
-            db.query(UpstreamGeneratedKey)
-            .filter(
-                UpstreamGeneratedKey.imported_website_id == wid,
-                UpstreamGeneratedKey.imported_account_id.isnot(None),
-                UpstreamGeneratedKey.status != "orphaned",
-            )
-            .all()
-        )
-        _backfill_missing_target_groups_from_remote_accounts(db, website, all_website_keys)
-
-        # ── 按竞争分组分桶 ──────────────────────────────────────────────────
-        # 竞争分组键：imported_target_group_id（老数据为 NULL 时 fallback 到 group_id）
-        buckets: dict[str, list[UpstreamGeneratedKey]] = defaultdict(list)
-        for row in all_website_keys:
-            comp_key = row.imported_target_group_id or row.group_id
-            buckets[comp_key].append(row)
-
-        # 只保留账号数 > 1 的分组（有竞争才需要排序）
-        competitive_buckets = {k: v for k, v in buckets.items() if len(v) > 1}
-
-        if not competitive_buckets:
-            logger.info(
-                "skip account priority sync for website %s: no competitive groups (all single-account)",
-                wid,
-            )
-            continue  # 不写日志，不发通知
-
-        # ── 预取快照倍率 ────────────────────────────────────────────────────
-        all_upstream_ids = {k.upstream_id for k in all_website_keys}
-        try:
-            # 构建 "{upstream_id}:{group_id}" → rate 查询表
-            raw_rate_map: dict[str, float] = {}
-            for uid in all_upstream_ids:
-                groups = latest_rate_map(db, uid)
-                for gid, g in groups.items():
-                    if isinstance(g, dict):
-                        raw_rate_map[f"{uid}:{gid}"] = _snapshot_group_rate(g)
-        except Exception as exc:
-            logger.warning("build rate map failed for website %s: %s", wid, exc)
-            raw_rate_map = {}
-
-        # ── 每个竞争分组内独立计算 priority ────────────────────────────────
-        # priority_assignment: account_id → new_priority
-        priority_assignment: dict[str, int] = {}
-        for comp_key, comp_rows in competitive_buckets.items():
-            # 只保留快照中能查到倍率的账号；无数据的账号不参与排序
-            rated = [
-                (row, raw_rate_map[f"{row.upstream_id}:{row.group_id}"])
-                for row in comp_rows
-                if f"{row.upstream_id}:{row.group_id}" in raw_rate_map
-            ]
-            # 过滤后有效账号不足 2 个 → 此分组无竞争意义，整组跳过
-            if len(rated) < 2:
-                logger.info(
-                    "skip competitive bucket %s for website %s: only %d account(s) have rate data",
-                    comp_key, wid, len(rated),
-                )
-                continue
-            # 组内按倍率升序排序（倍率低 → priority 小 → 优先）
-            unique_rates = sorted(set(r for _, r in rated))
-            rate_to_prio = {rate: priority_for_rate_rank(idx) for idx, rate in enumerate(unique_rates)}
-            for row, rate in rated:
-                priority_assignment[row.imported_account_id] = rate_to_prio[rate]
-
-        # ── 调用 update_account（仅竞争分组的账号）───────────────────────
+        # ── 1. 建立远端 client 连接并加载账号与分组列表 ────────────────────────
+        account_info_map: dict[str, dict[str, Any]] = {}
+        target_group_names: dict[str, str] = {}
+        remote_account_ids: set[str] | None = None
         site_results: list[dict] = []
+        competitive_buckets: dict[str, list] = {}
+        priority_assignment: dict[str, int] = {}
         try:
             with Sub2ApiWebsiteClient(
                 base_url=website.base_url,
@@ -548,18 +487,19 @@ def sync_account_priorities_for_upstream(
                 auth_config=json.loads(website.auth_config_json or "{}"),
                 timeout=float(website.timeout_seconds),
             ) as client:
-                # 尽量拉取目标网站账号列表以获取账号名称和旧 priority，并拉取分组以获取分组名称，失败不阻断
-                account_info_map: dict[str, dict[str, Any]] = {}
-                target_group_names: dict[str, str] = {}
                 try:
-                    remote_accounts = client.list_accounts() or []
-                    for acc in remote_accounts:
-                        acc_id = _extract_id(acc)
-                        if acc_id:
-                            account_info_map[str(acc_id)] = {
-                                "name": acc.get("name") or acc.get("account_name") or str(acc_id),
-                                "priority": acc.get("priority"),
-                            }
+                    remote_accounts = client.list_accounts()
+                    if remote_accounts is not None:
+                        remote_account_ids = set()
+                        for acc in remote_accounts:
+                            acc_id = _extract_id(acc)
+                            if acc_id:
+                                acc_id_str = str(acc_id)
+                                remote_account_ids.add(acc_id_str)
+                                account_info_map[acc_id_str] = {
+                                    "name": acc.get("name") or acc.get("account_name") or acc_id_str,
+                                    "priority": acc.get("priority"),
+                                }
                 except Exception as e:
                     logger.warning("failed to fetch remote accounts for website %s: %s", wid, e)
 
@@ -572,79 +512,201 @@ def sync_account_priorities_for_upstream(
                 except Exception as e:
                     logger.warning("failed to fetch groups for website %s at sync: %s", wid, e)
 
-                for comp_rows in competitive_buckets.values():
-                    for row in comp_rows:
-                        account_id = row.imported_account_id
-                        new_priority = priority_assignment.get(account_id)
-                        if new_priority is None:
-                            continue
-
-                        acc_info = account_info_map.get(str(account_id)) or {}
-                        acc_name = acc_info.get("name") or str(account_id)
-                        old_prio = acc_info.get("priority")
-                        src_rate = raw_rate_map.get(f"{row.upstream_id}:{row.group_id}")
+                # ── 2. 查询该网站所有已导入 Key（跨上游），用于倍率查询 ────────────
+                all_website_keys = (
+                    db.query(UpstreamGeneratedKey)
+                    .filter(
+                        UpstreamGeneratedKey.imported_website_id == wid,
+                        UpstreamGeneratedKey.imported_account_id.isnot(None),
+                        UpstreamGeneratedKey.status != "orphaned",
+                    )
+                    .all()
+                )
+                _backfill_missing_target_groups_from_remote_accounts(db, website, all_website_keys)
 
+                # ── 3. 自愈校验：清除远端已不存在的账号，并从后续计算中排除 ──────────
+                active_website_keys: list[UpstreamGeneratedKey] = []
+                for row in all_website_keys:
+                    account_id = row.imported_account_id
+                    if remote_account_ids is not None and str(account_id) not in remote_account_ids:
                         tg_id_str = str(row.imported_target_group_id or row.group_id)
-                        tg_name = target_group_names.get(tg_id_str)
-                        if tg_name and row.imported_target_group_name != tg_name:
-                            row.imported_target_group_name = tg_name
-                            db.commit()
+                        tg_name = target_group_names.get(tg_id_str) or row.imported_target_group_name
+                        site_results.append(
+                            _priority_result(
+                                row,
+                                None,
+                                "stale",
+                                "目标账号已不存在，已清除导入标记；请重新导入账号",
+                                account_name=account_id,
+                                old_priority=None,
+                                source_rate=None,
+                                target_group_name=tg_name,
+                            )
+                        )
+                        # 清理本地导入标记并恢复为 created
+                        row.imported_website_id = None
+                        row.imported_account_id = None
+                        row.imported_at = None
+                        row.status = "created"
+                        db.commit()
+                    else:
+                        active_website_keys.append(row)
 
-                        try:
-                            client.update_account(account_id, {"priority": new_priority})
+                # ── 4. 预取快照倍率 ────────────────────────────────────────────────
+                raw_rate_map: dict[str, float] = {}
+                if active_website_keys:
+                    all_upstream_ids = {k.upstream_id for k in active_website_keys}
+                    try:
+                        # 构建 "{upstream_id}:{group_id}" → rate 查询表
+                        for uid in all_upstream_ids:
+                            groups = latest_rate_map(db, uid)
+                            for gid, g in groups.items():
+                                if isinstance(g, dict):
+                                    raw_rate_map[f"{uid}:{gid}"] = _snapshot_group_rate(g)
+                    except Exception as exc:
+                        logger.warning("build rate map failed for website %s: %s", wid, exc)
+
+                # 补填 stale 账号结果的倍率
+                for res in site_results:
+                    if res["status"] == "stale":
+                        res["source_rate"] = raw_rate_map.get(f"{res['upstream_id']}:{res['group_id']}")
+
+                # ── 5. 基于仍存在的账号分桶并校验竞争 ──────────────────────────
+                buckets: dict[str, list[UpstreamGeneratedKey]] = defaultdict(list)
+                for row in active_website_keys:
+                    comp_key = row.imported_target_group_id or row.group_id
+                    buckets[comp_key].append(row)
+
+                # 只保留账号数 > 1 的分组（有竞争才需要排序）
+                competitive_buckets = {k: v for k, v in buckets.items() if len(v) > 1}
+
+                if not competitive_buckets:
+                    logger.info(
+                        "skip account priority sync for website %s: no competitive groups (all single-account)",
+                        wid,
+                    )
+                else:
+                    # ── 6. 每个竞争分组内独立计算 priority ──────────────────────
+                    priority_assignment: dict[str, int] = {}
+                    for comp_key, comp_rows in competitive_buckets.items():
+                        # 只保留快照中能查到倍率的账号；无数据的账号不参与排序
+                        rated = [
+                            (row, raw_rate_map[f"{row.upstream_id}:{row.group_id}"])
+                            for row in comp_rows
+                            if f"{row.upstream_id}:{row.group_id}" in raw_rate_map
+                        ]
+                        # 过滤后有效账号不足 2 个 → 此分组无竞争意义，整组跳过
+                        if len(rated) < 2:
                             logger.info(
-                                "updated priority for account %s (website=%s, upstream=%s, group=%s"
-                                ", comp_group=%s): %s",
-                                account_id, wid, row.upstream_id, row.group_id,
-                                row.imported_target_group_id or row.group_id, new_priority,
+                                "skip competitive bucket %s for website %s: only %d account(s) have rate data",
+                                comp_key, wid, len(rated),
                             )
-                            site_results.append(
-                                _priority_result(
-                                    row,
-                                    new_priority,
-                                    "success",
-                                    f"优先级已更新为 {new_priority}",
-                                    account_name=acc_name,
-                                    old_priority=old_prio,
-                                    source_rate=src_rate,
-                                    target_group_name=tg_name or row.imported_target_group_name,
+                            continue
+                        # 组内按倍率升序排序（倍率低 → priority 小 → 优先）
+                        unique_rates = sorted(set(r for _, r in rated))
+                        rate_to_prio = {rate: priority_for_rate_rank(idx) for idx, rate in enumerate(unique_rates)}
+                        for row, rate in rated:
+                            priority_assignment[row.imported_account_id] = rate_to_prio[rate]
+
+                    # ── 7. 更新竞争账号的 priority ──────────────────────────────
+                    for comp_rows in competitive_buckets.values():
+                        for row in comp_rows:
+                            account_id = row.imported_account_id
+                            new_priority = priority_assignment.get(account_id)
+                            if new_priority is None:
+                                continue
+
+                            acc_info = account_info_map.get(str(account_id)) or {}
+                            acc_name = acc_info.get("name") or str(account_id)
+                            old_prio = acc_info.get("priority")
+                            src_rate = raw_rate_map.get(f"{row.upstream_id}:{row.group_id}")
+
+                            tg_id_str = str(row.imported_target_group_id or row.group_id)
+                            tg_name = target_group_names.get(tg_id_str)
+                            if tg_name and row.imported_target_group_name != tg_name:
+                                row.imported_target_group_name = tg_name
+                                db.commit()
+
+                            try:
+                                client.update_account(account_id, {"priority": new_priority})
+                                logger.info(
+                                    "updated priority for account %s (website=%s, upstream=%s, group=%s"
+                                    ", comp_group=%s): %s",
+                                    account_id, wid, row.upstream_id, row.group_id,
+                                    row.imported_target_group_id or row.group_id, new_priority,
                                 )
-                            )
-                        except Exception as exc:
-                            logger.warning(
-                                "failed to update priority for account %s (website=%s): %s",
-                                account_id, wid, exc,
-                            )
-                            site_results.append(
-                                _priority_result(
-                                    row,
-                                    new_priority,
-                                    "failed",
-                                    str(exc),
-                                    account_name=acc_name,
-                                    old_priority=old_prio,
-                                    source_rate=src_rate,
-                                    target_group_name=tg_name or row.imported_target_group_name,
+                                site_results.append(
+                                    _priority_result(
+                                        row,
+                                        new_priority,
+                                        "success",
+                                        f"优先级已更新为 {new_priority}",
+                                        account_name=acc_name,
+                                        old_priority=old_prio,
+                                        source_rate=src_rate,
+                                        target_group_name=tg_name or row.imported_target_group_name,
+                                    )
+                                )
+                            except Exception as exc:
+                                # 兜底清理：如果 PUT 404
+                                import httpx
+                                is_404 = False
+                                if isinstance(exc, WebsiteError) and "不存在或已被删除" in str(exc):
+                                    is_404 = True
+                                elif isinstance(exc, WebsiteError) and isinstance(exc.__cause__, httpx.HTTPStatusError) and exc.__cause__.response.status_code == 404:
+                                    is_404 = True
+
+                                if is_404:
+                                    site_results.append(
+                                        _priority_result(
+                                            row,
+                                            new_priority,
+                                            "stale",
+                                            "目标账号已不存在，已清除导入标记；请重新导入账号",
+                                            account_name=acc_name,
+                                            old_priority=old_prio,
+                                            source_rate=src_rate,
+                                            target_group_name=tg_name or row.imported_target_group_name,
+                                        )
+                                    )
+                                    row.imported_website_id = None
+                                    row.imported_account_id = None
+                                    row.imported_at = None
+                                    row.status = "created"
+                                    db.commit()
+                                    continue
+
+                                logger.warning(
+                                    "failed to update priority for account %s (website=%s): %s",
+                                    account_id, wid, exc,
+                                )
+                                site_results.append(
+                                    _priority_result(
+                                        row,
+                                        new_priority,
+                                        "failed",
+                                        str(exc),
+                                        account_name=acc_name,
+                                        old_priority=old_prio,
+                                        source_rate=src_rate,
+                                        target_group_name=tg_name or row.imported_target_group_name,
+                                    )
                                 )
-                            )
         except Exception as exc:
             logger.warning("failed to connect website %s for account priority sync: %s", wid, exc)
-            for comp_rows in competitive_buckets.values():
-                for row in comp_rows:
-                    src_rate = raw_rate_map.get(f"{row.upstream_id}:{row.group_id}")
-                    site_results.append(
-                        _priority_result(
-                            row,
-                            None,
-                            "failed",
-                            f"连接网站失败: {exc}",
-                            source_rate=src_rate,
-                        )
+            for row in rows:
+                site_results.append(
+                    _priority_result(
+                        row,
+                        None,
+                        "failed",
+                        f"连接网站失败: {exc}",
                     )
+                )
 
-        # 只发送有实际成功/失败的通知（不包含单账号跳过项）
+        # ── 8. 写入日志与钉钉通知 ──────────────────────────────────────────
         notify_updates = [r for r in site_results if r["status"] in ("success", "failed")]
-        if notify_updates:
+        if notify_updates or any(r["status"] == "stale" for r in site_results):
             # 构建简化的 priority_map 供日志参考（只包含竞争分组的账号）
             priority_map_snapshot = {
                 f"{row.upstream_id}:{row.group_id}": priority_assignment.get(row.imported_account_id)
@@ -652,7 +714,8 @@ def sync_account_priorities_for_upstream(
                 for row in comp_rows
             }
             _write_priority_sync_log_with_map(db, wid, upstream_name, site_results, priority_map_snapshot)
-            _try_send_priority_webhook(db, wid, website.name, upstream_id, upstream_name, notify_updates)
+            if notify_updates:
+                _try_send_priority_webhook(db, wid, website.name, upstream_id, upstream_name, notify_updates)
 
         all_results.extend(site_results)
 
diff --git a/backend/test_priority_sync.py b/backend/test_priority_sync.py
index e4a08c2..f94b5c9 100644
--- a/backend/test_priority_sync.py
+++ b/backend/test_priority_sync.py
@@ -685,3 +685,194 @@ def test_priority_sync_graceful_on_remote_list_failure(db_session, monkeypatch):
     assert res_a1["old_priority"] is None
     assert res_a1["new_priority"] == 1
     assert res_a1["status"] == "success"
+
+
+def test_priority_sync_stale_account_cleanup_on_list_success(db_session, monkeypatch):
+    """验证 priority sync 时，若远端账号列表中缺失某个导入账号，能自愈清理本地标记并不影响其他竞争账号。"""
+    w = Website(name="W1", base_url="http://w1", enabled=True,
+                 auth_config_json="{}", timeout_seconds=30)
+    u = Upstream(name="U1", base_url="http://u1")
+    db_session.add_all([w, u])
+    db_session.commit()
+    db_session.refresh(w); db_session.refresh(u)
+
+    _make_snapshot(db_session, u.id, {"G1": 1.0, "G2": 2.0, "G3": 3.0})
+
+    # A1(K1), A2(K2) 与 A3(K3) 竞争
+    k1 = _make_key(db_session, u.id, "G1", "K1", "V1", w.id, "A1", imported_target_group_id="TG1")
+    k2 = _make_key(db_session, u.id, "G2", "K2", "V2", w.id, "A2", imported_target_group_id="TG1")
+    k3 = _make_key(db_session, u.id, "G3", "K3", "V3", w.id, "A3", imported_target_group_id="TG1")
+
+    update_calls = []
+
+    class MockClient:
+        def __init__(self, **kwargs): pass
+        def __enter__(self): return self
+        def __exit__(self, *a): pass
+        def list_accounts(self):
+            # 只有 A1 和 A2，缺失 A3
+            return [
+                {"id": "A1", "name": "SmartUp-A1"},
+                {"id": "A2", "name": "SmartUp-A2"},
+            ]
+        def get_groups(self, *a, **kw):
+            return [{"id": "TG1", "name": "TG1-Group"}]
+        def update_account(self, account_id, data):
+            update_calls.append((account_id, data))
+
+    monkeypatch.setattr("app.services.website_sync.Sub2ApiWebsiteClient", MockClient)
+
+    results = sync_account_priorities_for_upstream(db_session, u.id, website_id=w.id)
+
+    # A1 和 A2 正常竞争更新，A3 被自愈清除不调用 update_account
+    assert len(update_calls) == 2
+    priority_map = {aid: data["priority"] for aid, data in update_calls}
+    assert priority_map["A1"] == 1
+    assert priority_map["A2"] == 11
+
+    res_a3 = next(r for r in results if r["account_id"] == "A3")
+    assert res_a3["status"] == "stale"
+    assert "目标账号已不存在，已清除导入标记" in res_a3["message"]
+
+    db_session.refresh(k3)
+    assert k3.imported_account_id is None
+    assert k3.imported_website_id is None
+    assert k3.status == "created"
+
+
+def test_priority_sync_stale_account_single_remaining_skips_update(db_session, monkeypatch):
+    """验证 priority sync 时，若清除失效账号后某分组只剩一个有效账号，该分组跳过更新不调用 update_account。"""
+    w = Website(name="W1", base_url="http://w1", enabled=True,
+                 auth_config_json="{}", timeout_seconds=30)
+    u = Upstream(name="U1", base_url="http://u1")
+    db_session.add_all([w, u])
+    db_session.commit()
+    db_session.refresh(w); db_session.refresh(u)
+
+    _make_snapshot(db_session, u.id, {"G1": 1.0, "G2": 2.0})
+
+    # A1 与 A2 竞争
+    k1 = _make_key(db_session, u.id, "G1", "K1", "V1", w.id, "A1", imported_target_group_id="TG1")
+    k2 = _make_key(db_session, u.id, "G2", "K2", "V2", w.id, "A2", imported_target_group_id="TG1")
+
+    update_calls = []
+
+    class MockClient:
+        def __init__(self, **kwargs): pass
+        def __enter__(self): return self
+        def __exit__(self, *a): pass
+        def list_accounts(self):
+            # 只有 A1，缺失 A2
+            return [{"id": "A1", "name": "SmartUp-A1"}]
+        def get_groups(self, *a, **kw):
+            return [{"id": "TG1", "name": "TG1-Group"}]
+        def update_account(self, account_id, data):
+            update_calls.append((account_id, data))
+
+    monkeypatch.setattr("app.services.website_sync.Sub2ApiWebsiteClient", MockClient)
+
+    results = sync_account_priorities_for_upstream(db_session, u.id, website_id=w.id)
+
+    # 竞争分组只剩 A1 一个有效账号 -> 跳过不更新，update_account 不应被调用
+    assert update_calls == []
+    res_a2 = next(r for r in results if r["account_id"] == "A2")
+    assert res_a2["status"] == "stale"
+    db_session.refresh(k2)
+    assert k2.imported_account_id is None
+
+
+def test_priority_sync_no_cleanup_on_list_failure(db_session, monkeypatch):
+    """验证 list_accounts 失败时，不清理本地导入标记，仍尝试调用 update_account。"""
+    w = Website(name="W1", base_url="http://w1", enabled=True,
+                 auth_config_json="{}", timeout_seconds=30)
+    u = Upstream(name="U1", base_url="http://u1")
+    db_session.add_all([w, u])
+    db_session.commit()
+    db_session.refresh(w); db_session.refresh(u)
+
+    _make_snapshot(db_session, u.id, {"G1": 1.0, "G2": 2.0})
+
+    k1 = _make_key(db_session, u.id, "G1", "K1", "V1", w.id, "A1", imported_target_group_id="TG1")
+    k2 = _make_key(db_session, u.id, "G2", "K2", "V2", w.id, "A2", imported_target_group_id="TG1")
+
+    update_calls = []
+
+    class MockClient:
+        def __init__(self, **kwargs): pass
+        def __enter__(self): return self
+        def __exit__(self, *a): pass
+        def list_accounts(self):
+            raise Exception("Remote list error")
+        def update_account(self, account_id, data):
+            update_calls.append((account_id, data))
+
+    monkeypatch.setattr("app.services.website_sync.Sub2ApiWebsiteClient", MockClient)
+
+    sync_account_priorities_for_upstream(db_session, u.id, website_id=w.id)
+
+    # list_accounts 失败，不做清理，继续正常对 A1 和 A2 尝试更新
+    assert len(update_calls) == 2
+    db_session.refresh(k1)
+    db_session.refresh(k2)
+    assert k1.imported_account_id == "A1"
+    assert k2.imported_account_id == "A2"
+
+
+def test_update_account_404_error_custom_msg_and_fallback_cleanup(db_session, monkeypatch):
+    """验证 update_account 抛出 404 错误时转换为友好文案并触发自愈清除。"""
+    from app.services.website_client import WebsiteError
+    w = Website(name="W1", base_url="http://w1", enabled=True,
+                 auth_config_json="{}", timeout_seconds=30)
+    u = Upstream(name="U1", base_url="http://u1")
+    db_session.add_all([w, u])
+    db_session.commit()
+    db_session.refresh(w); db_session.refresh(u)
+
+    _make_snapshot(db_session, u.id, {"G1": 1.0, "G2": 2.0})
+
+    k1 = _make_key(db_session, u.id, "G1", "K1", "V1", w.id, "A1", imported_target_group_id="TG1")
+    k2 = _make_key(db_session, u.id, "G2", "K2", "V2", w.id, "A2", imported_target_group_id="TG1")
+
+    import httpx
+
+    class MockClient:
+        def __init__(self, **kwargs): pass
+        def __enter__(self): return self
+        def __exit__(self, *a): pass
+        def list_accounts(self):
+            # 模拟 list 失败，以便依赖 update_account 兜底清理
+            raise Exception("list failed")
+        def update_account(self, account_id, data):
+            # A2 触发 404 错误
+            if account_id == "A2":
+                req = httpx.Request("PUT", f"http://w1/accounts/{account_id}")
+                resp = httpx.Response(404, request=req)
+                raise WebsiteError("目标网站接口不存在") from httpx.HTTPStatusError("404", request=req, response=resp)
+
+        def get_groups(self, *a, **kw):
+            return []
+
+    # 验证单独调用 update_account 404 时的定制错误文案
+    client = Sub2ApiWebsiteClient(base_url="http://w1", api_prefix="", auth_type="api_key", auth_config={})
+    # Mock _request in client
+    def mock_request(method, path, body=None):
+        req = httpx.Request(method, path)
+        resp = httpx.Response(404, request=req)
+        raise WebsiteError("目标网站接口不存在") from httpx.HTTPStatusError("404", request=req, response=resp)
+    monkeypatch.setattr(client, "_request", mock_request)
+
+    with pytest.raises(WebsiteError) as exc_info:
+        client.update_account("A2", {"priority": 1})
+    assert "目标账号 A2 不存在或已被删除" in str(exc_info.value)
+
+    # 验证同步流程中的 404 兜底清理
+    monkeypatch.setattr("app.services.website_sync.Sub2ApiWebsiteClient", MockClient)
+    results = sync_account_priorities_for_upstream(db_session, u.id, website_id=w.id)
+
+    res_a2 = next(r for r in results if r["account_id"] == "A2")
+    assert res_a2["status"] == "stale"
+    assert "目标账号已不存在，已清除导入标记" in res_a2["message"]
+
+    db_session.refresh(k2)
+    assert k2.imported_account_id is None
+    assert k2.status == "created"