diff --git a/cytetype/preprocessing/validation.py b/cytetype/preprocessing/validation.py
index c785daf..d458848 100644
--- a/cytetype/preprocessing/validation.py
+++ b/cytetype/preprocessing/validation.py
@@ -28,7 +28,7 @@ def _is_gene_id_like(value: str) -> bool:
     if re.match(r"^[NX][MR]_\d+$", value):
         return True
 
-    if re.match(r"^\d+$", value):
+    if re.match(r"^\d+(?:\.0)?$", value):
         return True
 
     if re.match(r"^[A-Z0-9]+[._][A-Z0-9._]+$", value) and len(value) > 10:
@@ -102,7 +102,20 @@ def materialize_canonical_gene_symbols_column(
         source_name = f"column '{gene_symbols_column}'"
 
     canonical_column = _temporary_gene_symbols_column_name(adata)
-    adata.var[canonical_column] = clean_gene_names(source_values)
+    cleaned = clean_gene_names(source_values)
+
+    id_pct = _id_like_percentage(cleaned)
+    if id_pct > 49:
+        raise ValueError(
+            f"\n\nGene Symbol Detection Error\n"
+            f"{'─' * 50}\n"
+            f"CyteType requires human-readable gene symbols (e.g., TSPAN6, DPM1, SCYL3)\n"
+            f"To fix this, either:\n"
+            f"  1. Set gene_symbols_column to a column in adata.var that contains gene symbols\n"
+            f"  2. Convert your gene identifiers to symbols before running CyteType\n"
+        )
+
+    adata.var[canonical_column] = cleaned
     logger.info(
         f"Materialized canonical gene symbols in temporary column '{canonical_column}' "
         f"from {source_name}."
diff --git a/tests/test_validation.py b/tests/test_validation.py
new file mode 100644
index 0000000..b93e610
--- /dev/null
+++ b/tests/test_validation.py
@@ -0,0 +1,97 @@
+import pytest
+
+from cytetype.preprocessing.validation import _is_gene_id_like, _id_like_percentage
+
+
+class TestIsGeneIdLike:
+
+    @pytest.mark.parametrize("value", [
+        "ENSG00000000003",
+        "ENSG00000000003.14",
+        "ENSMUSG00000000001",
+        "ensg00000000003",
+    ])
+    def test_ensembl_ids(self, value: str) -> None:
+        assert _is_gene_id_like(value) is True
+
+    @pytest.mark.parametrize("value", [
+        "NM_001301",
+        "NR_046018",
+        "XM_011541",
+        "XR_001737",
+    ])
+    def test_refseq_ids(self, value: str) -> None:
+        assert _is_gene_id_like(value) is True
+
+    @pytest.mark.parametrize("value", [
+        "7157",
+        "672",
+        "11286",
+        "0",
+    ])
+    def test_integer_entrez_ids(self, value: str) -> None:
+        assert _is_gene_id_like(value) is True
+
+    @pytest.mark.parametrize("value", [
+        "7157.0",
+        "672.0",
+        "11286.0",
+        "0.0",
+    ])
+    def test_float_stringified_entrez_ids(self, value: str) -> None:
+        assert _is_gene_id_like(value) is True
+
+    @pytest.mark.parametrize("value", [
+        "AFFY_HG_U133A.207163_S_AT",
+        "ILLUMINA_HUMANHT_12_V4.ILMN_1762337",
+    ])
+    def test_long_dotted_ids(self, value: str) -> None:
+        assert _is_gene_id_like(value) is True
+
+    @pytest.mark.parametrize("value", [
+        "TSPAN6",
+        "DPM1",
+        "SCYL3",
+        "TP53",
+        "BRCA1",
+        "CD8A",
+        "MS4A1",
+    ])
+    def test_gene_symbols_not_flagged(self, value: str) -> None:
+        assert _is_gene_id_like(value) is False
+
+    @pytest.mark.parametrize("value", [
+        "",
+        "   ",
+        "7157.5",
+    ])
+    def test_edge_cases(self, value: str) -> None:
+        assert _is_gene_id_like(value) is False
+
+
+class TestIdLikePercentage:
+
+    def test_all_gene_symbols(self) -> None:
+        values = ["TSPAN6", "DPM1", "SCYL3", "TP53", "BRCA1"]
+        assert _id_like_percentage(values) == 0.0
+
+    def test_all_ensembl_ids(self) -> None:
+        values = [f"ENSG{i:011d}" for i in range(20)]
+        assert _id_like_percentage(values) == 100.0
+
+    def test_all_integer_entrez_ids(self) -> None:
+        values = ["7157", "672", "3845", "11286", "9952"]
+        assert _id_like_percentage(values) == 100.0
+
+    def test_all_float_entrez_ids(self) -> None:
+        values = ["7157.0", "672.0", "3845.0", "11286.0", "9952.0"]
+        assert _id_like_percentage(values) == 100.0
+
+    def test_mixed_float_entrez_and_symbols(self) -> None:
+        symbols = ["TSPAN6", "DPM1", "SCYL3"]
+        entrez = ["7157.0", "672.0", "3845.0", "11286.0", "9952.0", "904.0", "405.0"]
+        pct = _id_like_percentage(symbols + entrez)
+        assert pct == 70.0
+
+    def test_empty_list(self) -> None:
+        assert _id_like_percentage([]) == 100.0