diff --git a/backend/app/api/v1/endpoints/platform.py b/backend/app/api/v1/endpoints/platform.py
index f3387b3..34a77ba 100644
--- a/backend/app/api/v1/endpoints/platform.py
+++ b/backend/app/api/v1/endpoints/platform.py
@@ -2305,7 +2305,10 @@ async def model_eval_delete(task_id: str, current_user: dict = Depends(get_curre
     pending = _require_approval_or_admin("eval", task_id, current_user, f"删除评测任务 {task_id}")
     if pending:
         return pending
-    get_platform_store().delete_eval_task(task_id)
+    try:
+        get_platform_store().delete_eval_task(task_id)
+    except KeyError:
+        raise fail(404, "eval task not found")
     return ok({"deleted": task_id})
 
 
diff --git a/backend/app/db/platform_store.py b/backend/app/db/platform_store.py
index 2474354..ce35551 100644
--- a/backend/app/db/platform_store.py
+++ b/backend/app/db/platform_store.py
@@ -2515,12 +2515,18 @@ class PlatformStore:
 
     def eval_tasks(self) -> list[dict[str, Any]]:
         with self.connect() as conn:
-            rows = conn.execute("SELECT * FROM eval_tasks ORDER BY create_time DESC").fetchall()
+            # 评测任务采用软删除，普通列表不得再次返回已删除记录。
+            rows = conn.execute(
+                "SELECT * FROM eval_tasks WHERE deleted_at IS NULL ORDER BY create_time DESC"
+            ).fetchall()
             return [self._enrich_eval_payload(conn, self._json_payload_row(row)) for row in rows]
 
     def eval_task(self, task_id: str) -> dict[str, Any]:
         with self.connect() as conn:
-            row = conn.execute("SELECT * FROM eval_tasks WHERE id=?", (task_id,)).fetchone()
+            row = conn.execute(
+                "SELECT * FROM eval_tasks WHERE id=? AND deleted_at IS NULL",
+                (task_id,),
+            ).fetchone()
             if not row:
                 raise KeyError(task_id)
             payload = self._enrich_eval_payload(conn, self._json_payload_row(row))
@@ -2591,7 +2597,13 @@ class PlatformStore:
 
     def delete_eval_task(self, task_id: str) -> None:
         with self.connect() as conn:
-            conn.execute("UPDATE eval_tasks SET deleted_at=?, deleted_by=? WHERE id=?", (utcnow(), "system", task_id))
+            result = conn.execute(
+                "UPDATE eval_tasks SET deleted_at=?, deleted_by=? "
+                "WHERE id=? AND deleted_at IS NULL RETURNING id",
+                (utcnow(), "system", task_id),
+            )
+            if not result.fetchone():
+                raise KeyError(task_id)
 
     def running_eval_tasks(self) -> list[dict[str, Any]]:
         """Return eval tasks that have been submitted to a compute node and are still running."""