diff --git a/src/bytes/mod.rs b/src/bytes/mod.rs
index 96732cd57..a347a99a7 100644
--- a/src/bytes/mod.rs
+++ b/src/bytes/mod.rs
@@ -142,7 +142,14 @@ where
     let t = self.tag.clone();
 
     match i.compare_no_case(t) {
-      CompareResult::Ok => Ok((i.take_from(tag_len), OM::Output::bind(|| i.take(tag_len)))),
+      CompareResult::Ok => {
+        // Case folding can change a char's byte length (e.g. U+212A KELVIN SIGN folds
+        // to 'k'), so split on the input's element boundary, not the tag length.
+        let index = i
+          .slice_index(self.tag.iter_elements().count())
+          .unwrap_or(tag_len);
+        Ok((i.take_from(index), OM::Output::bind(|| i.take(index))))
+      }
       CompareResult::Incomplete => {
         if OM::Incomplete::is_streaming() {
           Err(Err::Incomplete(Needed::new(tag_len - i.input_len())))
diff --git a/tests/issues.rs b/tests/issues.rs
index 9996b5ccf..602d18a2d 100644
--- a/tests/issues.rs
+++ b/tests/issues.rs
@@ -281,3 +281,16 @@ fn issue_1808_complete_string_parser_returns_wrong_slice() {
     Ok(("", "\n"))
   );
 }
+
+#[test]
+fn issue_1883_tag_no_case_unicode_char_boundary() {
+  use nom::bytes::complete::tag_no_case;
+
+  // U+212A KELVIN SIGN (3 bytes) case-folds to ASCII 'k'; the split must land on a
+  // char boundary, and the returned slice is the input's own prefix.
+  let r: IResult<&str, &str> = tag_no_case("k")("\u{212A}xyz");
+  assert_eq!(r, Ok(("xyz", "\u{212A}")));
+
+  let r: IResult<&str, &str> = tag_no_case("aktz")("a\u{212A}tz!");
+  assert_eq!(r, Ok(("!", "a\u{212A}tz")));
+}