diff --git a/src/anndata/_io/specs/methods.py b/src/anndata/_io/specs/methods.py
index 5c84d83d8..91e11044b 100644
--- a/src/anndata/_io/specs/methods.py
+++ b/src/anndata/_io/specs/methods.py
@@ -541,6 +541,9 @@ def write_vlen_string_array_zarr(
         from numcodecs import VLenUTF8
         from zarr.codecs import VLenUTF8Codec
 
+        dataset_kwargs = dataset_kwargs.copy()
+        if "compressor" in dataset_kwargs:
+            dataset_kwargs.pop("compressor")
         f.create_array(
             k,
             shape=elem.shape,
@@ -1165,6 +1168,7 @@ def _remove_scalar_compression_args(dataset_kwargs: Mapping[str, Any]) -> dict:
         "shuffle",
         "fletcher32",
         "scaleoffset",
+        "compressor",
     ):
         dataset_kwargs.pop(arg, None)
     return dataset_kwargs
diff --git a/tests/test_readwrite.py b/tests/test_readwrite.py
index 50861ac6b..21cfca1fc 100644
--- a/tests/test_readwrite.py
+++ b/tests/test_readwrite.py
@@ -342,13 +342,14 @@ def test_zarr_compression(tmp_path):
 
     ad.io.write_zarr(pth, adata, compressor=compressor)
 
-    def check_compressed(key, value):
-        if isinstance(value, zarr.Array) and value.shape != ():
+    def check_compressed(value, key):
+        if value.shape != ():
             if value.compressor != compressor:
                 not_compressed.append(key)
 
-    with zarr.open(str(pth), "r") as f:
-        f.visititems(check_compressed)
+    f = zarr.open(str(pth), "r")
+    for key in f.array_keys():
+        check_compressed(f[key], key)
 
     if not_compressed:
         msg = "\n\t".join(not_compressed)