diff --git a/sqlite_utils/utils.py b/sqlite_utils/utils.py
index 91fd24ddd..0e631ddc2 100644
--- a/sqlite_utils/utils.py
+++ b/sqlite_utils/utils.py
@@ -9,6 +9,7 @@
 import json
 import os
 import sys
+from codecs import BOM_UTF8
 from collections.abc import Callable, Generator, Iterable, Iterator
 from typing import (
     TYPE_CHECKING,
@@ -382,7 +383,7 @@ class Format(enum.Enum):
         if not first_bytes:
             buffered.close()
             return (), Format.CSV
-        if first_bytes.startswith((b"[", b"{")):
+        if first_bytes.removeprefix(BOM_UTF8).lstrip().startswith((b"[", b"{")):
             # JSON is read eagerly, so the detection wrapper can close now,
             # including when parsing raises an error.
             # TODO: Detect newline-JSON
diff --git a/tests/test_rows_from_file.py b/tests/test_rows_from_file.py
index a8e7f9d76..4dc933b8c 100644
--- a/tests/test_rows_from_file.py
+++ b/tests/test_rows_from_file.py
@@ -1,5 +1,6 @@
 import io
 import json
+from codecs import BOM_UTF8
 from io import BytesIO, StringIO
 
 import pytest
@@ -13,6 +14,9 @@
         (b"id,name\n1,Cleo", Format.CSV),
         (b"id\tname\n1\tCleo", Format.TSV),
         (b'[{"id": "1", "name": "Cleo"}]', Format.JSON),
+        (BOM_UTF8 + b'[{"id": "1", "name": "Cleo"}]', Format.JSON),
+        (BOM_UTF8 + b'\n{"id": "1", "name": "Cleo"}', Format.JSON),
+        (BOM_UTF8 + b"id,name\n1,Cleo", Format.CSV),
     ),
 )
 def test_rows_from_file_detect_format(input, expected_format):