diff --git a/metaworld/envs/mujoco/sawyer_xyz/sawyer_xyz_env.py b/metaworld/envs/mujoco/sawyer_xyz/sawyer_xyz_env.py
index ee8f9114e..a0e9c1f5d 100644
--- a/metaworld/envs/mujoco/sawyer_xyz/sawyer_xyz_env.py
+++ b/metaworld/envs/mujoco/sawyer_xyz/sawyer_xyz_env.py
@@ -458,11 +458,9 @@ def sawyer_observation_space(self):
     def step(self, action):
         assert len(action) == 4, f"Actions should be size 4, got {len(action)}"
         self.set_xyz_action(action[:3])
+        if self.curr_path_length >= self.max_path_length:
+            raise ValueError("You must reset the env manually once truncate==True")
         self.do_simulation([action[-1], -action[-1]], n_frames=self.frame_skip)
-        if self.curr_path_length > self.max_path_length:
-            raise ValueError(
-                "Maximum path length allowed by the benchmark has been exceeded"
-            )
         self.curr_path_length += 1
 
         # Running the simulator can sometimes mess up site positions, so
@@ -496,12 +494,16 @@ def step(self, action):
             dtype=np.float64,
         )
         reward, info = self.evaluate_state(self._last_stable_obs, action)
-        # step will never return a terminal if there is a success
+        # step will never return a terminate==True if there is a success
+        # but we can return truncate=True if the current path length == max path length
+        truncate = False
+        if self.curr_path_length == self.max_path_length:
+            truncate = True
         return (
             np.array(self._last_stable_obs, dtype=np.float64),
             reward,
             False,
-            False,
+            truncate,
             info,
         )