diff --git a/sqlparse/engine/statement_splitter.py b/sqlparse/engine/statement_splitter.py
index bc57d170..5db9c9bf 100644
--- a/sqlparse/engine/statement_splitter.py
+++ b/sqlparse/engine/statement_splitter.py
@@ -182,7 +182,7 @@ def process(self, stream):
                 # Split on semicolon if not inside a BEGIN...END block
                 if self.level <= 0 and 'BEGIN' not in self._block_stack:
                     self.consume_ws = True
-            elif ttype is T.Keyword and value.split()[0] == 'GO':
+            elif ttype is T.Keyword and value.upper().split()[0] == 'GO':
                 self.consume_ws = True
             elif (ttype not in (T.Whitespace, T.Newline, T.Comment.Single,
                                 T.Comment.Multiline)
diff --git a/tests/test_split.py b/tests/test_split.py
index 92c3fefe..8a241c93 100644
--- a/tests/test_split.py
+++ b/tests/test_split.py
@@ -198,7 +198,12 @@ def test_split_strip_semicolon_procedure(load_file):
 @pytest.mark.parametrize('sql, num', [
     ('USE foo;\nGO\nSELECT 1;\nGO', 4),
     ('SELECT * FROM foo;\nGO', 2),
-    ('USE foo;\nGO 2\nSELECT 1;', 3)
+    ('USE foo;\nGO 2\nSELECT 1;', 3),
+    # issue762: the batch separator is case-insensitive
+    ('USE foo;\ngo\nSELECT 1;\ngo', 4),
+    ('USE foo;\nGo\nSELECT 1;\nGo', 4),
+    ('SELECT * FROM foo;\ngo', 2),
+    ('USE foo;\ngo 2\nSELECT 1;', 3),
 ])
 def test_split_go(sql, num):  # issue762
     stmts = sqlparse.split(sql)