diff --git a/DESCRIPTION.md b/DESCRIPTION.md
index 00f860ce..1ec1e072 100644
--- a/DESCRIPTION.md
+++ b/DESCRIPTION.md
@@ -9,7 +9,8 @@ Source code is also available at:
 
 # Release Notes
 - (Unreleased)
-  - Fix return value of snowflake get_table_names
+  - Fix return value of snowflake get_table_names.
+  - Fix incorrect quoting of identifiers with `_` as initial character.
   - Added `force_div_is_floordiv` flag to override `div_is_floordiv` new default value `False` in `SnowflakeDialect`.
     - With the flag in `False`, the `/` division operator will be treated as a float division and `//` as a floor division.
     - This flag is added to maintain backward compatibility with the previous behavior of Snowflake Dialect division.
diff --git a/src/snowflake/sqlalchemy/base.py b/src/snowflake/sqlalchemy/base.py
index 0226d37d..dc624949 100644
--- a/src/snowflake/sqlalchemy/base.py
+++ b/src/snowflake/sqlalchemy/base.py
@@ -117,7 +117,11 @@
     r"\s*(?:UPDATE|INSERT|DELETE|MERGE|COPY)", re.I | re.UNICODE
 )
 # used for quoting identifiers ie. table names, column names, etc.
-ILLEGAL_INITIAL_CHARACTERS = frozenset({d for d in string.digits}.union({"_", "$"}))
+ILLEGAL_INITIAL_CHARACTERS = frozenset({d for d in string.digits}.union({"$"}))
+
+
+# used for quoting identifiers ie. table names, column names, etc.
+ILLEGAL_IDENTIFIERS = frozenset({d for d in string.digits}.union({"_"}))
 
 """
 Overwrite methods to handle Snowflake BCR change:
@@ -443,6 +447,7 @@ def _join_left_to_right(
 class SnowflakeIdentifierPreparer(compiler.IdentifierPreparer):
     reserved_words = {x.lower() for x in RESERVED_WORDS}
     illegal_initial_characters = ILLEGAL_INITIAL_CHARACTERS
+    illegal_identifiers = ILLEGAL_IDENTIFIERS
 
     def __init__(self, dialect, **kw):
         quote = '"'
@@ -471,6 +476,17 @@ def format_label(self, label, name=None):
 
         return self.quote_identifier(s) if n.quote else s
 
+    def _requires_quotes(self, value: str) -> bool:
+        """Return True if the given identifier requires quoting."""
+        lc_value = value.lower()
+        return (
+            lc_value in self.reserved_words
+            or lc_value in self.illegal_identifiers
+            or value[0] in self.illegal_initial_characters
+            or not self.legal_characters.match(str(value))
+            or (lc_value != value)
+        )
+
     def _split_schema_by_dot(self, schema):
         ret = []
         idx = 0
diff --git a/tests/test_compiler.py b/tests/test_compiler.py
index 0eea4607..cb9632a4 100644
--- a/tests/test_compiler.py
+++ b/tests/test_compiler.py
@@ -50,6 +50,21 @@ def test_underscore_as_valid_identifier(self):
             dialect="snowflake",
         )
 
+    def test_underscore_as_initial_character_as_non_quoted_identifier(self):
+        _table = table(
+            "table_1745924",
+            column("ca", Integer),
+            column("cb", String),
+            column("_identifier", String),
+        )
+
+        stmt = insert(_table).values(ca=1, cb="test", _identifier="test_")
+        self.assert_compile(
+            stmt,
+            "INSERT INTO table_1745924 (ca, cb, _identifier) VALUES (%(ca)s, %(cb)s, %(_identifier)s)",
+            dialect="snowflake",
+        )
+
     def test_multi_table_delete(self):
         statement = table1.delete().where(table1.c.id == table2.c.id)
         self.assert_compile(