diff --git a/tests/test_parser.py b/tests/test_parser.py
index 375932e..338eb0d 100644
--- a/tests/test_parser.py
+++ b/tests/test_parser.py
@@ -73,6 +73,23 @@ def test_malformed_array_still_raises() -> None:
             Parser(content).parse()
 
 
+def test_array_with_malformed_element_raises() -> None:
+    # An element that starts to parse like a value but then hits an
+    # unexpected char (e.g. an inline table missing its "=" or "}") must
+    # surface the error instead of being swallowed and silently dropped,
+    # which left the array truncated to whatever parsed before it.
+    for content in (
+        "a = [{1]",
+        "a = [{a]",
+        "a = [1, {2]",
+        "a = [{a = 1}, {2]",
+        "a = [[1, {x]]",
+        "a = [{a.b]",
+    ):
+        with pytest.raises(UnexpectedCharError):
+            Parser(content).parse()
+
+
 def test_parse_multiline_string_ignore_the_first_newline() -> None:
     content = 'a = """\nfoo\n"""'
     parser = Parser(content)
diff --git a/tomlkit/parser.py b/tomlkit/parser.py
index b952b36..165bb2e 100644
--- a/tomlkit/parser.py
+++ b/tomlkit/parser.py
@@ -665,12 +665,9 @@ def _parse_array(self) -> Array:
             # exception eagerly computes a line/column, which scans the whole
             # source. On a large file with many arrays this is a big, pure waste.
             if not prev_value and self._current != "]":
-                try:
-                    elems.append(self._parse_value())
-                    prev_value = True
-                    continue
-                except UnexpectedCharError:
-                    pass
+                elems.append(self._parse_value())
+                prev_value = True
+                continue
 
             # consume comma
             if prev_value and self._current == ",":