diff --git a/vllm/v1/worker/gpu_worker.py b/vllm/v1/worker/gpu_worker.py
index 58e2d658c42b..6d117175b1ad 100644
--- a/vllm/v1/worker/gpu_worker.py
+++ b/vllm/v1/worker/gpu_worker.py
@@ -392,8 +392,10 @@ def determine_available_memory(self) -> int:
             )
 
             # Profile CUDA graph memory if graphs will be captured.
+            # Skip on ROCm/HIP as graph pool handles and mem_get_info behave
+            # differently and can produce incorrect/negative estimates.
             cudagraph_memory_estimate = 0
-            if not self.model_config.enforce_eager:
+            if not self.model_config.enforce_eager and not current_platform.is_rocm():
                 cudagraph_memory_estimate = self.model_runner.profile_cudagraph_memory()
 
         # Use the pre-cudagraph torch peak to avoid double-counting.
@@ -406,6 +408,8 @@ def determine_available_memory(self) -> int:
             + profile_result.weights_memory
         )
 
+        # On ROCm, cudagraph_memory_estimate is always 0 so this is a no-op.
+        # On CUDA, respect the opt-in flag as originally designed.
         cudagraph_memory_estimate_applied = (
             cudagraph_memory_estimate
             if envs.VLLM_MEMORY_PROFILER_ESTIMATE_CUDAGRAPHS
@@ -517,7 +521,6 @@ def get_kv_cache_spec(self) -> dict[str, KVCacheSpec]:
 
     def update_max_model_len(self, max_model_len: int) -> None:
         """Update max_model_len after auto-fit to GPU memory.
-
         This is called when max_model_len=-1 is used and the engine
         automatically determines the maximum context length that fits
         in GPU memory. Workers need to update their cached max_model_len