diff --git a/paddle/fluid/operators/scatter.h b/paddle/fluid/operators/scatter.h
index 802d9ae76d3b5894100b8f86c979f6b4fb12a161..c4f9c628dbb04ebba860931ad9f2ca3b11a3473e 100644
--- a/paddle/fluid/operators/scatter.h
+++ b/paddle/fluid/operators/scatter.h
@@ -169,15 +169,9 @@ void ScatterAssignAdd(const framework::ExecutionContext& ctx, const Tensor& src,
   const size_t& slice_bytes = slice_size * sizeof(T);
 
   // if not in overwrite mode, need to init output data
+  auto max_index = dst_dims[0];
   for (int64_t i = 0; i < index_size; ++i) {
     const IndexT& index_val = p_index[i];
-    memset(p_output + slice_size * index_val, 0, slice_bytes);
-  }
-
-  // if not in overwrite mode, need to init output data
-  for (int64_t i = 0; i < index_size; ++i) {
-    const IndexT& index_val = p_index[i];
-
     PADDLE_ENFORCE_GE(index_val, 0,
                       platform::errors::OutOfRange(
                           "The index is out of bounds, "
@@ -185,7 +179,19 @@ void ScatterAssignAdd(const framework::ExecutionContext& ctx, const Tensor& src,
                           "input meet the requirements. It should "
                           "be greater than or equal to 0, but received [%d]",
                           index_val));
+    PADDLE_ENFORCE_LT(index_val, max_index,
+                      platform::errors::OutOfRange(
+                          "The index is out of bounds, "
+                          "please check whether the dimensions of index and "
+                          "input meet the requirements. It should "
+                          "be less than %d, but received %d",
+                          max_index, index_val));
+    memset(p_output + slice_size * index_val, 0, slice_bytes);
+  }
 
+  // if not in overwrite mode, need to init output data
+  for (int64_t i = 0; i < index_size; ++i) {
+    const IndexT& index_val = p_index[i];
     elementwise_inner_add<T, IndexT>(ctx, p_src, p_output, i, index_val,
                                      slice_size);
   }