diff --git a/CHANGELOG.md b/CHANGELOG.md
index e31b8d15..df6beb20 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -6,6 +6,13 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 This project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.html),
 with the exception that 0.x versions can break between minor versions.
 
+## Unreleased
+### Fixed
+- Fix parsing of link reference definitions where it looks like it has a title
+  but it doesn't because it's followed by characters other than space/tab. In that
+  case, the title was set to the partially-parsed title and the source spans were
+  wrong (#315).
+
 ## [0.22.0] - 2024-03-15
 ### Added
 - New `MarkdownRenderer` for rendering nodes to Markdown (CommonMark)!
diff --git a/commonmark/src/main/java/org/commonmark/internal/LinkReferenceDefinitionParser.java b/commonmark/src/main/java/org/commonmark/internal/LinkReferenceDefinitionParser.java
index ebe6ebb2..b58e669e 100644
--- a/commonmark/src/main/java/org/commonmark/internal/LinkReferenceDefinitionParser.java
+++ b/commonmark/src/main/java/org/commonmark/internal/LinkReferenceDefinitionParser.java
@@ -15,7 +15,7 @@
 /**
  * Parser for link reference definitions at the beginning of a paragraph.
  *
- * @see <a href="https://spec.commonmark.org/0.29/#link-reference-definition">Link reference definitions</a>
+ * @see <a href="https://spec.commonmark.org/0.31.2/#link-reference-definitions">Link reference definitions</a>
  */
 public class LinkReferenceDefinitionParser {
 
@@ -70,6 +70,9 @@ public void parse(SourceLine line) {
             // Parsing failed, which means we fall back to treating text as a paragraph.
             if (!success) {
                 state = State.PARAGRAPH;
+                // If parsing of the title part failed, we still have a valid reference that we can add, and we need to
+                // do it before the source span for this line is added.
+                finishReference();
                 return;
             }
         }
@@ -218,7 +221,8 @@ private boolean startTitle(Scanner scanner) {
     private boolean title(Scanner scanner) {
         Position start = scanner.position();
         if (!LinkScanner.scanLinkTitleContent(scanner, titleDelimiter)) {
-            // Invalid title, stop
+            // Invalid title, stop. Title collected so far must not be used.
+            title = null;
             return false;
         }
 
@@ -235,6 +239,8 @@ private boolean title(Scanner scanner) {
         scanner.whitespace();
         if (scanner.hasNext()) {
             // spec: No further non-whitespace characters may occur on the line.
+            // Title collected so far must not be used.
+            title = null;
             return false;
         }
         referenceValid = true;
diff --git a/commonmark/src/test/java/org/commonmark/internal/LinkReferenceDefinitionParserTest.java b/commonmark/src/test/java/org/commonmark/internal/LinkReferenceDefinitionParserTest.java
index b4f57739..3f22adac 100644
--- a/commonmark/src/test/java/org/commonmark/internal/LinkReferenceDefinitionParserTest.java
+++ b/commonmark/src/test/java/org/commonmark/internal/LinkReferenceDefinitionParserTest.java
@@ -145,6 +145,30 @@ public void testTitleMultiline2() {
         assertDef(parser.getDefinitions().get(0), "foo", "/url", "\ntitle");
     }
 
+    @Test
+    public void testTitleMultiline3() {
+        parse("[foo]: /url");
+        assertEquals(State.START_TITLE, parser.getState());
+        // Note that this looks like a valid title until we parse "bad", at which point we need to treat the whole line
+        // as a paragraph line and discard any already parsed title.
+        parse("\"title\" bad");
+        assertEquals(State.PARAGRAPH, parser.getState());
+
+        assertDef(parser.getDefinitions().get(0), "foo", "/url", null);
+    }
+
+    @Test
+    public void testTitleMultiline4() {
+        parse("[foo]: /url");
+        assertEquals(State.START_TITLE, parser.getState());
+        parse("(title");
+        assertEquals(State.TITLE, parser.getState());
+        parse("foo(");
+        assertEquals(State.PARAGRAPH, parser.getState());
+
+        assertDef(parser.getDefinitions().get(0), "foo", "/url", null);
+    }
+
     @Test
     public void testTitleInvalid() {
         assertParagraph("[foo]: /url (invalid(");
diff --git a/commonmark/src/test/java/org/commonmark/test/SourceSpansTest.java b/commonmark/src/test/java/org/commonmark/test/SourceSpansTest.java
index 95959893..59241e49 100644
--- a/commonmark/src/test/java/org/commonmark/test/SourceSpansTest.java
+++ b/commonmark/src/test/java/org/commonmark/test/SourceSpansTest.java
@@ -183,6 +183,15 @@ public void linkReferenceDefinitionWithTitle() {
         assertEquals(List.of(SourceSpan.of(1, 0, 11)), def2.getSourceSpans());
     }
 
+    @Test
+    public void linkReferenceDefinitionWithTitleInvalid() {
+        var doc = PARSER.parse("[foo]: /url\n\"title\" ok\n");
+        var def = Nodes.find(doc, LinkReferenceDefinition.class);
+        var paragraph = Nodes.find(doc, Paragraph.class);
+        assertEquals(List.of(SourceSpan.of(0, 0, 11)), def.getSourceSpans());
+        assertEquals(List.of(SourceSpan.of(1, 0, 10)), paragraph.getSourceSpans());
+    }
+
     @Test
     public void linkReferenceDefinitionHeading() {
         // This is probably the trickiest because we have a link reference definition at the start of a paragraph