diff --git a/pairtools/lib/scaling.py b/pairtools/lib/scaling.py
index 772c163..7746c43 100644
--- a/pairtools/lib/scaling.py
+++ b/pairtools/lib/scaling.py
@@ -389,7 +389,11 @@ def compute_scaling(
         pairs_df = pairs
 
     elif isinstance(pairs, str) or hasattr(pairs, "buffer") or hasattr(pairs, "peek"):
-        pairs_df, _, _ = pairsio.read_pairs(pairs, nproc=nproc_in, chunksize=chunksize)
+        pairs_df, _, header_chromsizes = pairsio.read_pairs(
+            pairs, nproc=nproc_in, chunksize=chunksize
+        )
+        if chromsizes is None and header_chromsizes is not None and len(header_chromsizes):
+            chromsizes = header_chromsizes
     else:
         raise ValueError(
             "pairs must be either a path to a pairs file or a pd.DataFrame"
diff --git a/tests/test_scaling.py b/tests/test_scaling.py
index 2104600..225ba83 100644
--- a/tests/test_scaling.py
+++ b/tests/test_scaling.py
@@ -25,3 +25,16 @@ def test_scaling():
     assert (
         output["n_pairs"].sum() == 9
     )  # double unmapped pairs are currently ignored by lib.scaling
+
+
+def test_scaling_uses_header_chromsizes():
+    # Without an explicit view, the implicit per-chromosome view must use the chromsizes
+    # from the .pairs header (chr* = 100 in mock.pairsam); otherwise region ends stay at
+    # -1 and n_bp2 is left empty.
+    from pairtools.lib.scaling import compute_scaling
+
+    mock_pairsam_path = os.path.join(testdir, "data", "mock.pairsam")
+    cis, _trans = compute_scaling(mock_pairsam_path, regions=None)
+
+    assert (cis["end1"] == 100).all()
+    assert (cis["n_bp2"] > 0).any()