diff --git a/refactor/ast.py b/refactor/ast.py
index 3d23262..d1a5c4f 100644
--- a/refactor/ast.py
+++ b/refactor/ast.py
@@ -139,6 +139,7 @@ class PreciseUnparser(BaseUnparser):
     for major statements and child node recovery."""
 
     def __init__(self, *args: Any, **kwargs: Any) -> None:
+        self.empty_lines = kwargs.pop("empty_lines", None)
         self._visited_comment_lines: set[int] = set()
         super().__init__(*args, **kwargs)
 
@@ -222,10 +223,12 @@ def _write_if_unseen_comment(
         preceding_comments = []
         for offset, line in enumerate(reversed(lines[:node_start])):
             comment_begin = line.find("#")
-            if comment_begin == -1 or comment_begin != node.col_offset:
+            not_valid_comment: bool = comment_begin == -1 or comment_begin != node.col_offset
+            no_empty_lines: bool = not self.empty_lines or line and not line.isspace()
+            if no_empty_lines and not_valid_comment:
                 break
 
-            preceding_comments.append((node_start - offset, line, comment_begin))
+            preceding_comments.append((node_start - offset, line, node.col_offset))
 
         for comment_info in reversed(preceding_comments):
             _write_if_unseen_comment(*comment_info)
@@ -240,7 +243,7 @@ def _write_if_unseen_comment(
             _write_if_unseen_comment(
                 line_no=node_end + offset,
                 line=line,
-                comment_begin=comment_begin,
+                comment_begin=node.col_offset,
             )
 
     def collect_comments(self, node: ast.AST) -> ContextManager[None]:
diff --git a/tests/test_ast.py b/tests/test_ast.py
index 1e5dc1d..b0eddf7 100644
--- a/tests/test_ast.py
+++ b/tests/test_ast.py
@@ -69,7 +69,7 @@ def test_split_lines_with_encoding(case):
         else:
             start_line = lines[lineno][col_offset:]
             end_line = lines[end_lineno][:end_col_offset]
-            match = start_line + lines[lineno + 1 : end_lineno].join() + end_line
+            match = start_line + lines[lineno + 1: end_lineno].join() + end_line
 
         assert str(match) == ast.get_source_segment(case, node)
 
@@ -240,3 +240,113 @@ def foo():
 
     base = PreciseUnparser(source=source)
     assert base.unparse(tree) + "\n" == expected_src
+
+
+def test_precise_empty_lines_unparser():
+    source = textwrap.dedent(
+        """\
+    def func():
+        if something:
+            print(
+                call(.1),
+                maybe+something_else,
+                maybe / other,
+                thing   . a
+            )
+    """
+    )
+
+    expected_src = textwrap.dedent(
+        """\
+    def func():
+        if something:
+            print(call(.1), maybe+something_else, maybe / other, thing   . a, 3)
+    """
+    )
+
+    tree = ast.parse(source)
+    tree.body[0].body[0].body[0].value.args.append(ast.Constant(3))
+
+    base = PreciseUnparser(source=source, empty_lines=True)
+    assert base.unparse(tree) + "\n" == expected_src
+
+
+def test_precise_empty_lines_unparser_indented_literals():
+    source = textwrap.dedent(
+        """\
+    def func():
+        if something:
+            print(
+                "bleh"
+                "zoom"
+            )
+    """
+    )
+
+    expected_src = textwrap.dedent(
+        """\
+    def func():
+        if something:
+            print("bleh"
+                "zoom", 3)
+    """
+    )
+
+    tree = ast.parse(source)
+    tree.body[0].body[0].body[0].value.args.append(ast.Constant(3))
+
+    base = PreciseUnparser(source=source, empty_lines=True)
+    assert base.unparse(tree) + "\n" == expected_src
+
+
+def test_precise_empty_lines_unparser_comments():
+    source = textwrap.dedent(
+        """\
+def foo():
+# unindented comment
+    # indented but not connected comment
+    
+    # a
+    # a1
+    print()
+    # a2
+    print()
+    # b
+    
+    # b2
+    print(
+        c # e
+    )
+    # c
+    print(d)
+    # final comment
+"""
+    )
+
+    expected_src = (
+        """\
+def foo():
+    # indented but not connected comment
+    
+    # a
+    # a1
+    print()
+    # a2
+    print()
+    # b
+    
+    # b2
+    print(
+        c # e
+    )
+    # c
+"""
+    )
+
+    tree = ast.parse(source)
+
+    # # Remove the print(d)
+    tree.body[0].body.pop()
+
+    base = PreciseUnparser(source=source, empty_lines=True)
+    assert base.unparse(tree) + "\n" == expected_src