diff --git a/osm4gpd/parse.py b/osm4gpd/parse.py
index f312359..bd6a965 100644
--- a/osm4gpd/parse.py
+++ b/osm4gpd/parse.py
@@ -65,6 +65,11 @@ class OSMFile:
     ways: list[WayGroup] = field(default_factory=list)
     relations: list[RelationGroup] = field(default_factory=list)
 
+    # protected property that is used to store the arguments to filter
+    # for later use during consolidation, since pre-consolidation filtering
+    # can leave non-matching geometries that are referenced by some way or relation
+    _filter: set[str] | None = None
+
     @classmethod
     def from_file(cls, fp: Path | str) -> OSMFile:
         if isinstance(fp, str):
@@ -97,17 +102,20 @@ def filter(self, *, tags: set[str]) -> OSMFile:
             self.ways, tags=tags, references=references
         )
         self.nodes, _ = filter_groups(self.nodes, tags=tags, references=references)
+
+        self._filter = tags
         return self
 
-    def consolidate(self) -> gpd.GeoDataFrame:
+    def _consolidate_nodes(self) -> gpd.GeoDataFrame:
         _node_parts = [
             consolidate_nodes(nodes) for nodes in self.nodes if not nodes.is_empty()
         ]
         if len(_node_parts) > 0:
-            nodes = pd.concat(_node_parts)
+            return pd.concat(_node_parts)
         else:
             raise ValueError("Nothing to consolidate.")
 
+    def _consolidate_ways(self, *, nodes: gpd.GeoDataFrame) -> gpd.GeoDataFrame:
         _way_parts = [
             consolidate_ways(ways, nodes=nodes)
             for ways in self.ways
@@ -115,10 +123,13 @@ def consolidate(self) -> gpd.GeoDataFrame:
         ]
 
         if len(_way_parts) > 0:
-            ways = pd.concat(_way_parts)
+            return pd.concat(_way_parts)
         else:
-            ways = gpd.GeoDataFrame()
+            return gpd.GeoDataFrame()
 
+    def _consolidate_relations(
+        self, *, nodes: gpd.GeoDataFrame, ways: gpd.GeoDataFrame
+    ) -> gpd.GeoDataFrame:
         _relation_parts = [
             consolidate_relations(relations, ways=ways, nodes=nodes)
             for relations in self.relations
@@ -126,10 +137,25 @@ def consolidate(self) -> gpd.GeoDataFrame:
         ]
 
         if len(_relation_parts) > 0:
-            relations = pd.concat(_relation_parts)
-            return pd.concat([nodes, ways, relations])
+            return pd.concat(_relation_parts)
         else:
-            return pd.concat([nodes, ways])
+            return gpd.GeoDataFrame()
+
+    def consolidate(self) -> gpd.GeoDataFrame:
+        nodes = self._consolidate_nodes()
+        ways = self._consolidate_ways(nodes=nodes)
+        relations = self._consolidate_relations(nodes=nodes, ways=ways)
+
+        gdf = pd.concat([nodes, ways, relations])
+
+        if self._filter is not None:
+            # filter for rows that match a filter category
+            gdf = gdf[gdf[list(self._filter)].notna().any(axis=1)]
+
+            # drop columns that became all NA from filtering
+            gdf = gdf[gdf.columns[~gdf.isna().all()]]
+
+        return gdf
 
 
 #  header_bbox = box(
diff --git a/tests/test_consolidate.py b/tests/test_consolidate.py
index 91deba2..f229ff1 100644
--- a/tests/test_consolidate.py
+++ b/tests/test_consolidate.py
@@ -6,13 +6,13 @@
 @pytest.mark.parametrize(
     "filename,tags,expected_shape",
     [
-        ("extract", {"amenity"}, (1184, 157)),
-        ("isle_of_man", {"name"}, (102488, 1109)),
-        ("malta", {"amenity"}, (19247, 354)),
-        ("malta", {"car_wash"}, (10, 8)),
+        ("extract", {"amenity"}, (314, 155)),
+        ("isle_of_man", {"name"}, (10827, 1062)),
+        ("malta", {"amenity"}, (5396, 342)),
+        ("malta", {"car_wash"}, (1, 8)),
     ],
 )
-def test_osm_file_can_be_consolidated(
+def test_osm_file_can_be_consolidated_after_filtering(
     filename: str,
     tags: set[str],
     expected_shape: tuple[int, int],
@@ -25,3 +25,40 @@ def test_osm_file_can_be_consolidated(
     )
 
     assert gdf.shape == expected_shape
+
+
+@pytest.mark.parametrize(
+    "filename,expected_shape",
+    [
+        ("extract", (1184, 157)),
+    ],
+)
+def test_osm_file_can_be_consolidated_without_filtering(
+    filename: str,
+    expected_shape: tuple[int, int],
+    request: pytest.FixtureRequest,
+) -> None:
+    gdf = OSMFile.from_file(request.getfixturevalue(filename)).consolidate()
+
+    assert gdf.shape == expected_shape
+
+
+@pytest.mark.parametrize(
+    "filename,tags",
+    [
+        ("extract", {"amenity"}),
+        ("malta", {"car_wash"}),
+    ],
+)
+def test_consolidation_respects_filtering(
+    filename: str,
+    tags: set[str],
+    request: pytest.FixtureRequest,
+) -> None:
+    gdf = (
+        OSMFile.from_file(request.getfixturevalue(filename))
+        .filter(tags=tags)
+        .consolidate()
+    )
+
+    assert gdf[list(tags)].isna().sum().sum() == 0