diff --git a/src/qs_codec/decode.py b/src/qs_codec/decode.py
index 0d75e47..8275746 100644
--- a/src/qs_codec/decode.py
+++ b/src/qs_codec/decode.py
@@ -137,7 +137,8 @@ def _parse_array_value(value: t.Any, options: DecodeOptions, current_list_length
     - When ``list_limit`` is negative:
         * if ``raise_on_limit_exceeded=True``, **any** list-growth operation here (e.g., comma-splitting)
           raises immediately;
-        * if ``raise_on_limit_exceeded=False`` (default), comma-splitting still returns a list; numeric
+        * if ``raise_on_limit_exceeded=False`` (default), comma-splitting is allowed here and is normalized
+          later in ``_parse_query_string_values`` (which applies overflow conversion when needed). Numeric
           bracket indices are handled later by ``_parse_object`` (where negative ``list_limit`` disables
           numeric-index parsing only).
 
@@ -176,7 +177,10 @@ def _parse_query_string_values(value: str, options: DecodeOptions) -> t.Dict[str
         * Decode key/value via ``options.decoder`` (default: percent-decoding using the selected ``charset``).
           Keys are passed with ``kind=DecodeKind.KEY`` and values with ``kind=DecodeKind.VALUE``; a custom decoder
           may return the raw token or ``None``.
-        * Apply comma-split list logic to values (handled here). Index-based list growth from bracket segments is applied later in ``_parse_object``. When ``list_limit < 0`` and ``raise_on_limit_exceeded=True``, any comma-split that would increase the list length raises immediately; otherwise the split proceeds.
+        * Apply comma-split list logic to values (handled here). Index-based list growth from bracket segments
+          is applied later in ``_parse_object``. When ``comma=True`` yields a list longer than ``list_limit``,
+          the value either raises (``raise_on_limit_exceeded=True``) or is normalized through overflow conversion
+          (``raise_on_limit_exceeded=False``), mirroring indexed/bracket list-limit handling.
         * Interpret numeric entities for Latin-1 when requested.
         * Handle empty brackets ``[]`` as list markers (wrapping exactly once).
         * Merge duplicate keys according to ``duplicates`` policy.
@@ -285,6 +289,14 @@ def _parse_query_string_values(value: str, options: DecodeOptions) -> t.Dict[str
         if options.parse_lists and "[]=" in part:
             val = [val]
 
+        # Enforce comma-split list limits consistently with other list construction paths.
+        if options.comma and isinstance(val, list) and len(val) > options.list_limit:
+            if options.raise_on_limit_exceeded:
+                raise ValueError(
+                    f"List limit exceeded: Only {options.list_limit} element{'' if options.list_limit == 1 else 's'} allowed in a list."
+                )
+            val = Utils.combine([], val, options)
+
         existing: bool = key in obj
 
         # Combine/overwrite according to the configured duplicates policy.
diff --git a/tests/unit/decode_test.py b/tests/unit/decode_test.py
index c2175a1..9cac89f 100644
--- a/tests/unit/decode_test.py
+++ b/tests/unit/decode_test.py
@@ -1370,6 +1370,75 @@ def test_list_limit(
         else:
             assert decode(query, options) == expected
 
+    @pytest.mark.parametrize(
+        "query, options, expected, raises_error, expect_overflow",
+        [
+            pytest.param(
+                "a=1,2,3",
+                DecodeOptions(comma=True, list_limit=5),
+                {"a": ["1", "2", "3"]},
+                False,
+                False,
+                id="comma-within-list-limit",
+            ),
+            pytest.param(
+                "a=1,2,3,4",
+                DecodeOptions(comma=True, list_limit=3),
+                {"a": {"0": "1", "1": "2", "2": "3", "3": "4"}},
+                False,
+                True,
+                id="comma-over-list-limit-converts-to-overflow",
+            ),
+            pytest.param(
+                "a=1,2,3,4",
+                DecodeOptions(comma=True, list_limit=3, raise_on_limit_exceeded=True),
+                None,
+                True,
+                False,
+                id="comma-over-list-limit-raises",
+            ),
+            pytest.param(
+                "a=1,2,3",
+                DecodeOptions(comma=True, list_limit=3),
+                {"a": ["1", "2", "3"]},
+                False,
+                False,
+                id="comma-at-list-limit-stays-list",
+            ),
+        ],
+    )
+    def test_comma_list_limit_parity(
+        self,
+        query: str,
+        options: DecodeOptions,
+        expected: t.Optional[t.Mapping[str, t.Any]],
+        raises_error: bool,
+        expect_overflow: bool,
+    ) -> None:
+        if raises_error:
+            with pytest.raises(ValueError, match="List limit exceeded"):
+                decode(query, options)
+            return
+
+        result = decode(query, options)
+        assert result == expected
+        if expect_overflow:
+            assert isinstance(result["a"], OverflowDict)
+        else:
+            assert isinstance(result["a"], list)
+
+    def test_comma_list_limit_raises_when_decoder_returns_oversized_list(self) -> None:
+        def _decoder(s: t.Optional[str], charset: t.Optional[Charset], *, kind: DecodeKind = DecodeKind.VALUE) -> t.Any:
+            if kind is DecodeKind.VALUE and s == "1":
+                return ["x", "y"]
+            return DecodeUtils.decode(s, charset=charset, kind=kind)
+
+        with pytest.raises(ValueError, match="List limit exceeded"):
+            decode(
+                "a=1",
+                DecodeOptions(comma=True, list_limit=1, raise_on_limit_exceeded=True, decoder=_decoder),
+            )
+
 
 # --- Additional tests for decoder kind and parser state isolation ---