diff --git a/src/laptrack/_cost_matrix.py b/src/laptrack/_cost_matrix.py
index 3959c7e0..4a4ff35c 100644
--- a/src/laptrack/_cost_matrix.py
+++ b/src/laptrack/_cost_matrix.py
@@ -43,12 +43,20 @@ def build_frame_cost_matrix(
 
     if track_start_cost is None:
         if len(C.data) > 0:
-            track_start_cost = np.max(C.data) * 1.05
+            max_val = np.max(C.data)
+            if max_val > 0:
+                track_start_cost = max_val * 1.05
+            else:
+                track_start_cost = EPSILON
         else:
             track_start_cost = 1.05
     if track_end_cost is None:
         if len(C.data) > 0:
-            track_end_cost = np.max(C.data) * 1.05
+            max_val = np.max(C.data)
+            if max_val > 0:
+                track_end_cost = max_val * 1.05
+            else:
+                track_end_cost = EPSILON
         else:
             track_end_cost = 1.05
 
@@ -159,6 +167,8 @@ def build_segment_cost_matrix(
             )
             * alternative_cost_factor
         )
+        if alternative_cost == 0:
+            alternative_cost = EPSILON
         if track_start_cost is None:
             track_start_cost = alternative_cost
         if track_end_cost is None:
diff --git a/tests/data/same_position_example.csv b/tests/data/same_position_example.csv
new file mode 100644
index 00000000..7a43c6a4
--- /dev/null
+++ b/tests/data/same_position_example.csv
@@ -0,0 +1,76 @@
+,seconds,h
+0,0,0
+1,1,0
+2,2,3
+3,3,5
+4,4,6
+5,5,6
+6,6,8
+7,7,9
+8,8,11
+9,9,12
+10,10,13
+11,11,14
+12,12,15
+13,13,16
+14,14,17
+15,15,19
+16,16,20
+17,17,20
+18,18,23
+19,19,23
+20,20,25
+21,21,27
+22,22,28
+23,23,29
+24,24,29
+25,25,0
+26,25,30
+27,26,0
+28,26,31
+29,27,3
+30,27,32
+31,28,5
+32,28,34
+33,29,7
+34,29,34
+35,30,8
+36,30,35
+37,31,9
+38,31,35
+39,32,12
+40,32,36
+41,33,13
+42,33,38
+43,34,14
+44,34,39
+45,35,15
+46,35,41
+47,36,16
+48,36,42
+49,37,17
+50,37,43
+51,38,19
+52,38,44
+53,39,20
+54,39,45
+55,40,21
+56,40,46
+57,41,22
+58,41,47
+59,42,22
+60,42,48
+61,43,23
+62,43,50
+63,44,24
+64,44,50
+65,45,25
+66,45,51
+67,46,26
+68,46,52
+69,47,27
+70,47,54
+71,48,28
+72,48,54
+73,49,28
+74,49,56
diff --git a/tests/test_overlap_tracking.py b/tests/test_overlap_tracking.py
index e2f0a623..d0745624 100644
--- a/tests/test_overlap_tracking.py
+++ b/tests/test_overlap_tracking.py
@@ -123,9 +123,13 @@ def metric(c1, c2, params):
     track_df2, split_df2, merge_df2 = olt.predict_overlap_dataframe(labels)
 
     track_df1 = track_df1.droplevel("index").set_index(["label"], append=True)
-    assert all(track_df1[["tree_id", "track_id"]] == track_df2[["tree_id", "track_id"]])
-    assert all(split_df1 == split_df2)
-    assert all(merge_df1 == merge_df2)
+    assert (
+        (track_df1[["tree_id", "track_id"]] == track_df2[["tree_id", "track_id"]])
+        .all()
+        .all()
+    )
+    assert (split_df1 == split_df2).all().all()
+    assert (merge_df1 == merge_df2).all().all()
 
 
 """
diff --git a/tests/test_tracking.py b/tests/test_tracking.py
index 0db9c56a..b1cd2363 100644
--- a/tests/test_tracking.py
+++ b/tests/test_tracking.py
@@ -133,9 +133,9 @@ def test_reproducing_trackmate(testdata, parallel_backend) -> None:
     assert not any(merge_df.duplicated())
     track_df2, split_df2, merge_df2 = convert_tree_to_dataframe(track_tree, coords)
     track_df2 = track_df2.rename(columns={"coord-0": "x", "coord-1": "y"})
-    assert all(track_df == track_df2)
-    assert all(split_df == split_df2)
-    assert all(merge_df == merge_df2)
+    assert (track_df == track_df2).all().all()
+    assert (split_df == split_df2).all().all()
+    assert (merge_df == merge_df2).all().all()
 
     # check index offset
     track_df3, split_df3, merge_df3 = lt.predict_dataframe(
@@ -155,18 +155,18 @@ def test_reproducing_trackmate(testdata, parallel_backend) -> None:
     if not merge_df4.empty:
         merge_df4["parent_track_id"] = merge_df4["parent_track_id"] + 2
         merge_df4["child_track_id"] = merge_df4["child_track_id"] + 2
-    assert all(track_df3 == track_df2)
-    assert all(split_df3 == split_df2)
-    assert all(merge_df3 == merge_df2)
+    assert (track_df3 == track_df4).all().all()
+    assert (split_df3 == split_df4).all().all()
+    assert (merge_df3 == merge_df4).all().all()
 
     track_df, split_df, merge_df = lt.predict_dataframe(
         df, ["x", "y"], only_coordinate_cols=False
     )
     assert all(track_df["frame_y"] == track_df2.index.get_level_values("frame"))
     track_df = track_df.drop(columns=["frame_y"])
-    assert all(track_df == track_df2)
-    assert all(split_df == split_df2)
-    assert all(merge_df == merge_df2)
+    assert (track_df == track_df2).all().all()
+    assert (split_df == split_df2).all().all()
+    assert (merge_df == merge_df2).all().all()
 
 
 @pytest.fixture(params=[2, 3, 4])
@@ -199,6 +199,37 @@ def test_tracking_zero_distance() -> None:
     assert set(edges) == set([((0, 0), (1, 0)), ((0, 1), (1, 1))])
 
 
+def test_tracking_zero_distance2(shared_datadir: str) -> None:
+    data = pd.read_csv(path.join(shared_datadir, "same_position_example.csv"))
+    lt = LapTrack(
+        track_cost_cutoff=15**2,
+        splitting_cost_cutoff=15**2,
+        merging_cost_cutoff=15**2,
+    )
+
+    track_df1, split_df1, merge_df1 = lt.predict_dataframe(
+        data,
+        coordinate_cols=["h"],
+        frame_col="seconds",
+        only_coordinate_cols=False,
+    )
+
+    data2 = data.copy()
+    data2["h"] += np.random.random(len(data2)) * 1e-2
+    track_df2, split_df2, merge_df2 = lt.predict_dataframe(
+        data2,
+        coordinate_cols=["h"],
+        frame_col="seconds",
+        only_coordinate_cols=False,
+    )
+
+    assert (
+        (track_df1[["track_id", "tree_id"]] == track_df2[["track_id", "tree_id"]])
+        .all()
+        .all()
+    )
+
+
 def test_tracking_not_connected() -> None:
     coords = [np.array([[10, 10], [12, 11]]), np.array([[50, 50], [53, 51]])]
     lt = LapTrack(