diff --git a/docs/history.md b/docs/history.md
index 9adb7cef..c5567005 100644
--- a/docs/history.md
+++ b/docs/history.md
@@ -2,6 +2,7 @@
 
 ## Development
 
+* FIX: ensure `to_cfradial2` correctly selects the default storage engine when none is provided, ({pull}`378`) by [@chfer](https://github.com/chfer)
 * MNT: Add ``cfradial1_sgp_file`` session fixture and refactor 8 tests in ``test_util.py``/``test_accessors.py`` to share it instead of inlining ``DATASETS.fetch("sample_sgp_data.nc")``. Fixture returns the filename so each test opens its own DataTree, avoiding cross-test mutation ({issue}`346`, {pull}`347`) by [@aladinor](https://github.com/aladinor)
 * FIX: IRIS reader rotates the first-loaded moment in each sweep by 1 ray — ``IrisRawFile._get_ray_record_offsets_and_data`` initialised ``j = -1`` so the first matching ray of the first-loaded moment was written to ``raw_data[-1]``; affects files without ``DB_XHDR`` (data-type bit 0) where ``DB_DBT`` becomes the rotated moment ({issue}`357`, {pull}`375`) by [@aladinor](https://github.com/aladinor)
 * DOC: Add projection comparison and cartopy map examples to ``Georeference_TargetCRS`` notebook by [@syedhamidali](https://github.com/syedhamidali)
diff --git a/tests/io/test_cfradial2.py b/tests/io/test_cfradial2.py
index 698a68ee..8fa92c72 100644
--- a/tests/io/test_cfradial2.py
+++ b/tests/io/test_cfradial2.py
@@ -8,6 +8,7 @@
 
 import xradar as xd
 from xradar.io.backends import cfradial2 as cf2
+from xradar.io.export import cfradial2 as export_cf2
 
 
 def _write_institutional_cfradial2(outfile):
@@ -138,6 +139,37 @@ def test_open_cfradial2_invalid_path():
         xd.io.open_cfradial2_datatree("missing-cfradial2-file.nc")
 
 
+@pytest.mark.parametrize(
+    ("available", "expected_engine"),
+    [
+        ("netCDF4", "netcdf4"),
+        ("h5netcdf", "h5netcdf"),
+    ],
+)
+def test_to_cfradial2_selects_default_engine(
+    monkeypatch, tmp_path, available, expected_engine
+):
+    # build a tiny DataTree with the minimum structure needed for export, and a root history attribute since export appends to it
+    dtree = xr.DataTree.from_dict({"/": xr.Dataset(attrs={"history": ""})})
+
+    # monkeypatch has_import to simulate different available engines, and ensure that the expected one is selected when engine=None
+    monkeypatch.setattr(export_cf2, "has_import", lambda name: name == available)
+
+    # monkeypatch DataTree.to_netcdf to avoid writing a real file and let the test intercept the `engine` argument
+    seen = {}
+
+    def _capture_to_netcdf(self, filename, engine=None, **kwargs):
+        seen["engine"] = engine
+
+    monkeypatch.setattr(xr.DataTree, "to_netcdf", _capture_to_netcdf, raising=True)
+
+    # call to_cfradial2 with engine=None, which should trigger the default engine selection logic
+    export_cf2.to_cfradial2(dtree, tmp_path / "out.nc", engine=None)
+
+    # does the intercepted call to to_netcdf have the expected engine ?
+    assert seen["engine"] == expected_engine
+
+
 @pytest.mark.parametrize(
     ("name", "expected"),
     [
diff --git a/xradar/io/export/cfradial2.py b/xradar/io/export/cfradial2.py
index 019aa7d6..66ab8b1b 100644
--- a/xradar/io/export/cfradial2.py
+++ b/xradar/io/export/cfradial2.py
@@ -58,9 +58,9 @@ def to_cfradial2(dtree, filename, engine=None, timestep=None):
     """
     if engine is None:
         if has_import("netCDF4"):
-            engine == "netcdf4"
+            engine = "netcdf4"
         elif has_import("h5netcdf"):
-            engine == "h5netcdf"
+            engine = "h5netcdf"
         else:
             raise ImportError(
                 "xradar: ``netCDF4`` or ``h5netcdf`` needed to perform this operation."