diff --git a/paddle/fluid/operators/tensorrt/tensorrt_engine_op.h b/paddle/fluid/operators/tensorrt/tensorrt_engine_op.h
index b1aa2b2c49ef6d8ef174f5294491accced4faa89..b0ac285b5d38d63a1205e589b6593298777b30a3 100644
--- a/paddle/fluid/operators/tensorrt/tensorrt_engine_op.h
+++ b/paddle/fluid/operators/tensorrt/tensorrt_engine_op.h
@@ -467,13 +467,6 @@ class TensorRTEngineOp : public framework::OperatorBase {
     auto stream =
         reinterpret_cast<const platform::CUDADeviceContext &>(dev_ctx).stream();
 
-    PADDLE_ENFORCE_EQ(
-        runtime_input_names_.empty(),
-        false,
-        platform::errors::PreconditionNotMet(
-            "TensorRT engine needs at least one input, but no input is found. "
-            "Please check if you set the input correctly."));
-
     std::vector<std::string> output_maps =
         Attr<std::vector<std::string>>("output_name_mapping");