diff --git a/library/core/src/main/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParser.java b/library/core/src/main/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParser.java
index 280ec4112b..b82fd93b03 100644
--- a/library/core/src/main/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParser.java
+++ b/library/core/src/main/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParser.java
@@ -48,6 +48,7 @@ import java.lang.annotation.Retention;
 import java.util.ArrayDeque;
 import java.util.ArrayList;
 import java.util.Collections;
+import java.util.Comparator;
 import java.util.HashMap;
 import java.util.List;
 import java.util.Map;
@@ -532,27 +533,7 @@ public final class WebvttCueParser {
             Spanned.SPAN_EXCLUSIVE_EXCLUSIVE);
         break;
       case TAG_RUBY:
-        @Nullable Element rubyTextElement = null;
-        for (int i = 0; i < nestedElements.size(); i++) {
-          if (TAG_RUBY_TEXT.equals(nestedElements.get(i).startTag.name)) {
-            rubyTextElement = nestedElements.get(i);
-            // Behaviour of multiple <rt> tags inside <ruby> is undefined, so use the first one.
-            break;
-          }
-        }
-        if (rubyTextElement == null) {
-          break;
-        }
-        // Move the rubyText from spannedText into the RubySpan.
-        CharSequence rubyText =
-            text.subSequence(rubyTextElement.startTag.position, rubyTextElement.endPosition);
-        text.delete(rubyTextElement.startTag.position, rubyTextElement.endPosition);
-        end -= rubyText.length();
-        text.setSpan(
-            new RubySpan(rubyText.toString(), RubySpan.POSITION_OVER),
-            start,
-            end,
-            Spanned.SPAN_EXCLUSIVE_EXCLUSIVE);
+        applyRubySpans(nestedElements, text, start);
         break;
       case TAG_UNDERLINE:
         text.setSpan(new UnderlineSpan(), start, end, Spanned.SPAN_EXCLUSIVE_EXCLUSIVE);
@@ -575,6 +556,34 @@ public final class WebvttCueParser {
     }
   }
 
+  private static void applyRubySpans(
+      List<Element> nestedElements, SpannableStringBuilder text, int startTagPosition) {
+    List<Element> sortedNestedElements = new ArrayList<>(nestedElements.size());
+    sortedNestedElements.addAll(nestedElements);
+    Collections.sort(sortedNestedElements, Element.BY_START_POSITION_ASC);
+    int deletedCharCount = 0;
+    int lastRubyTextEnd = startTagPosition;
+    for (int i = 0; i < sortedNestedElements.size(); i++) {
+      if (!TAG_RUBY_TEXT.equals(sortedNestedElements.get(i).startTag.name)) {
+        continue;
+      }
+      Element rubyTextElement = sortedNestedElements.get(i);
+      // Move the rubyText from spannedText into the RubySpan.
+      int adjustedRubyTextStart = rubyTextElement.startTag.position - deletedCharCount;
+      int adjustedRubyTextEnd = rubyTextElement.endPosition - deletedCharCount;
+      CharSequence rubyText = text.subSequence(adjustedRubyTextStart, adjustedRubyTextEnd);
+      text.delete(adjustedRubyTextStart, adjustedRubyTextEnd);
+      text.setSpan(
+          new RubySpan(rubyText.toString(), RubySpan.POSITION_OVER),
+          lastRubyTextEnd,
+          adjustedRubyTextStart,
+          Spanned.SPAN_EXCLUSIVE_EXCLUSIVE);
+      deletedCharCount += rubyText.length();
+      // The ruby text has been deleted, so new-start == old-end.
+      lastRubyTextEnd = adjustedRubyTextStart;
+    }
+  }
+
   /**
    * Adds {@link ForegroundColorSpan}s and {@link BackgroundColorSpan}s to {@code text} for entries
    * in {@code classes} that match WebVTT's <a
@@ -921,6 +930,9 @@ public final class WebvttCueParser {
 
   /** Information about a complete element (i.e. start tag and end position). */
   private static class Element {
+    private static final Comparator<Element> BY_START_POSITION_ASC =
+        (e1, e2) -> Integer.compare(e1.startTag.position, e2.startTag.position);
+
     private final StartTag startTag;
     /**
      * The position of the end of this element's text in the un-marked-up cue text (i.e. the
diff --git a/library/core/src/test/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParserTest.java b/library/core/src/test/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParserTest.java
index 36d052ed6c..f500029885 100644
--- a/library/core/src/test/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParserTest.java
+++ b/library/core/src/test/java/com/google/android/exoplayer2/text/webvtt/WebvttCueParserTest.java
@@ -62,6 +62,29 @@ public final class WebvttCueParserTest {
         .withTextAndPosition("with ruby", RubySpan.POSITION_OVER);
   }
 
+  @Test
+  public void parseSingleRubyTagWithMultipleRts() throws Exception {
+    Spanned text = parseCueText("<ruby>A<rt>1</rt>B<rt>2</rt>C<rt>3</rt></ruby>");
+
+    // The text between the <rt> tags is stripped from Cue.text and only present on the RubySpan.
+    assertThat(text.toString()).isEqualTo("ABC");
+    assertThat(text).hasRubySpanBetween(0, 1).withTextAndPosition("1", RubySpan.POSITION_OVER);
+    assertThat(text).hasRubySpanBetween(1, 2).withTextAndPosition("2", RubySpan.POSITION_OVER);
+    assertThat(text).hasRubySpanBetween(2, 3).withTextAndPosition("3", RubySpan.POSITION_OVER);
+  }
+
+  @Test
+  public void parseMultipleRubyTagsWithSingleRtEach() throws Exception {
+    Spanned text =
+        parseCueText("<ruby>A<rt>1</rt></ruby><ruby>B<rt>2</rt></ruby><ruby>C<rt>3</rt></ruby>");
+
+    // The text between the <rt> tags is stripped from Cue.text and only present on the RubySpan.
+    assertThat(text.toString()).isEqualTo("ABC");
+    assertThat(text).hasRubySpanBetween(0, 1).withTextAndPosition("1", RubySpan.POSITION_OVER);
+    assertThat(text).hasRubySpanBetween(1, 2).withTextAndPosition("2", RubySpan.POSITION_OVER);
+    assertThat(text).hasRubySpanBetween(2, 3).withTextAndPosition("3", RubySpan.POSITION_OVER);
+  }
+
   @Test
   public void parseRubyTagWithNoTextTag() throws Exception {
     Spanned text = parseCueText("Some <ruby>base text with no ruby text</ruby>");