diff --git a/tensorrt_llm/_torch/pyexecutor/py_executor.py b/tensorrt_llm/_torch/pyexecutor/py_executor.py
index 6cf98b759f02..78642af31d28 100644
--- a/tensorrt_llm/_torch/pyexecutor/py_executor.py
+++ b/tensorrt_llm/_torch/pyexecutor/py_executor.py
@@ -4947,8 +4947,13 @@ def _fetch_and_enqueue_requests(self, waiting_queue: WaitingQueue,
         if len(self.control_requests) != 0:
             return
 
-        # Calculate timeout
-        idle = (total_num_active_requests == 0) and len(waiting_queue) == 0
+        # Calculate timeout. Never wait once the shutdown sentinel has been
+        # consumed: no further item will arrive to wake a blocking get(), so
+        # blocking would keep the loop from reaching the
+        # `should_stop_processing` check that ends it, deadlocking shutdown()
+        # on `shutdown_event`.
+        idle = (total_num_active_requests == 0 and len(waiting_queue) == 0
+                and not self.is_shutdown)
         if idle:
             # In Ray path (TLLM_DISABLE_MPI=1), use a periodic heartbeat timeout so rank 0
             # reaches the broadcast path regularly to prevent trtllm-serve timeout when idle.