diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7e2768f2..90b46723 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -118,6 +118,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   users on other CUDA versions install the matching `torch` variant manually
   (see README) [\#604](https://github.com/mllam/neural-lam/pull/604) @RajdeepKushwaha5
 
+- Add edge count consistency check to `test_graph_creation.py` [#301](https://github.com/mllam/neural-lam/pull/301) @osten-antonio
+
 ## [v0.6.0](https://github.com/mllam/neural-lam/releases/tag/v0.6.0)
 
 This release introduces new features including GIF animation support, wandb run resumption, and improved ensemble loading, alongside a large number of bug fixes and maintenance updates.
diff --git a/tests/test_graph_creation.py b/tests/test_graph_creation.py
index 3eca73a2..e076c31f 100644
--- a/tests/test_graph_creation.py
+++ b/tests/test_graph_creation.py
@@ -51,6 +51,14 @@ def test_graph_creation(datastore_name, graph_name):
         "m2g_features.pt",
         "mesh_features.pt",
     ]
+
+    # index-feature pair to check if edge is consistent across files
+    edge_index_feature_pairs = [
+        ("g2m_edge_index", "g2m_features"),
+        ("m2g_edge_index", "m2g_features"),
+        ("m2m_edge_index", "m2m_features"),
+    ]
+
     if hierarchical:
         required_graph_files.extend(
             [
@@ -60,11 +68,17 @@ def test_graph_creation(datastore_name, graph_name):
                 "mesh_down_features.pt",
             ]
         )
+        edge_index_feature_pairs.extend(
+            [
+                ("mesh_up_edge_index", "mesh_up_features"),
+                ("mesh_down_edge_index", "mesh_down_features"),
+            ]
+        )
 
-    # TODO: check that the number of edges is consistent over the files, for
-    # now we just check the number of features
+    # check that the number of edges is consistent over the files
     d_features = 3
     d_mesh_static = 2
+    edge_counts = {}
 
     with tempfile.TemporaryDirectory() as tmpdir:
         graph_dir_path = Path(tmpdir) / "graph" / graph_name
@@ -75,7 +89,6 @@ def test_graph_creation(datastore_name, graph_name):
             hierarchical=hierarchical,
             n_max_levels=n_max_levels,
         )
-
         assert graph_dir_path.exists()
 
         # check that all the required files are present
@@ -89,13 +102,14 @@ def test_graph_creation(datastore_name, graph_name):
 
             if file_id.startswith("g2m") or file_id.startswith("m2g"):
                 assert isinstance(result, torch.Tensor)
-
                 if file_id.endswith("_index"):
                     assert (
                         result.shape[0] == 2
                     )  # adjacency matrix uses two rows
+                    edge_counts[file_id] = result.shape[1]
                 elif file_id.endswith("_features"):
                     assert result.shape[1] == d_features
+                    edge_counts[file_id] = result.shape[0]
 
             elif file_id.startswith("m2m") or file_id.startswith("mesh"):
                 assert isinstance(result, list)
@@ -119,6 +133,20 @@ def test_graph_creation(datastore_name, graph_name):
                     elif file_id.endswith("_features"):
                         assert r.shape[1] == d_features
 
+                if file_id.endswith("_index"):
+                    edge_counts[file_id] = [r.shape[1] for r in result]
+                elif (
+                    file_id.endswith("_features") and file_id != "mesh_features"
+                ):
+                    edge_counts[file_id] = [r.shape[0] for r in result]
+
+    # loop through index-feature pair to check consistency
+    for index_id, features_id in edge_index_feature_pairs:
+        assert edge_counts[index_id] == edge_counts[features_id], (
+            f"Edge count mismatch: {index_id} has {edge_counts[index_id]} edges"
+            f" but {features_id} has {edge_counts[features_id]} rows"
+        )
+
 
 @pytest.mark.parametrize("graph_name", ["1level", "multiscale", "hierarchical"])
 @pytest.mark.parametrize("datastore_name", DATASTORES.keys())