diff --git a/packages/netlify-cms-widget-markdown/src/serializers/__tests__/remarkEscapeMarkdownEntities.spec.js b/packages/netlify-cms-widget-markdown/src/serializers/__tests__/remarkEscapeMarkdownEntities.spec.js
index 86372ea4..a8336d14 100644
--- a/packages/netlify-cms-widget-markdown/src/serializers/__tests__/remarkEscapeMarkdownEntities.spec.js
+++ b/packages/netlify-cms-widget-markdown/src/serializers/__tests__/remarkEscapeMarkdownEntities.spec.js
@@ -73,7 +73,13 @@ describe('remarkEscapeMarkdownEntities', () => {
     expect(process('a b <pre>*c*</pre> d e')).toEqual('a b <pre>*c*</pre> d e');
   });
 
-  it('should not parse footnotes', () => {
-    expect(process('[^a]')).toEqual('\\[^a]');
+  it('should not escape footnote references', () => {
+    expect(process('[^a]')).toEqual('[^a]');
+    expect(process('[^1]')).toEqual('[^1]');
+  });
+
+  it('should not escape footnotes', () => {
+    expect(process('[^a]:')).toEqual('[^a]:');
+    expect(process('[^1]:')).toEqual('[^1]:');
   });
 });
diff --git a/packages/netlify-cms-widget-markdown/src/serializers/remarkEscapeMarkdownEntities.js b/packages/netlify-cms-widget-markdown/src/serializers/remarkEscapeMarkdownEntities.js
index a9e0609b..8c696a48 100644
--- a/packages/netlify-cms-widget-markdown/src/serializers/remarkEscapeMarkdownEntities.js
+++ b/packages/netlify-cms-widget-markdown/src/serializers/remarkEscapeMarkdownEntities.js
@@ -124,14 +124,14 @@ const escapePatterns = [
   /(`+)[^`]*(\1)/g,
 
   /**
-   * Links, Images, References, and Footnotes
+   * Links and Images
    *
-   * Match strings surrounded by brackets. This could be improved to
-   * specifically match only the exact syntax of each covered entity, but
-   * doing so through current approach would incur a considerable performance
-   * penalty.
+   * Match strings surrounded by square brackets, except when the opening
+   * bracket is followed by a caret. This could be improved to specifically
+   * match only the exact syntax of each covered entity, but doing so through
+   * current approach would incur a considerable performance penalty.
    */
-  /(\[)[^\]]*]/g,
+  /(\[(?!\^)+)[^\]]*]/g,
 ];
 
 /**