diff --git a/lib/banzai/filter/gollum_tags_filter.rb b/lib/banzai/filter/gollum_tags_filter.rb
index fe01dae4850..bcf5297e382 100644
--- a/lib/banzai/filter/gollum_tags_filter.rb
+++ b/lib/banzai/filter/gollum_tags_filter.rb
@@ -50,21 +50,30 @@ module Banzai
       # See https://github.com/gollum/gollum/wiki
       #
       # Rubular: http://rubular.com/r/7dQnE5CUCH
-      TAGS_PATTERN = %r{\[\[(.+?)\]\]}
+      TAGS_PATTERN = %r{\[\[(.+?)\]\]}.freeze
 
       # Pattern to match allowed image extensions
-      ALLOWED_IMAGE_EXTENSIONS = %r{.+(jpg|png|gif|svg|bmp)\z}i
+      ALLOWED_IMAGE_EXTENSIONS = %r{.+(jpg|png|gif|svg|bmp)\z}i.freeze
 
       def call
         search_text_nodes(doc).each do |node|
-          content = node.content
+          # A Gollum ToC tag is `[[_TOC_]]`, but due to MarkdownFilter running
+          # before this one, it will be converted into `[[<em>TOC</em>]]`, so it
+          # needs special-case handling
+          if toc_tag?(node)
+            next unless result[:toc].present?
 
-          next unless content.match(TAGS_PATTERN)
+            process_toc_tag(node)
+          else
+            content = node.content
 
-          html = process_tag($1)
+            next unless content =~ TAGS_PATTERN
 
-          if html && html != node.content
-            node.replace(html)
+            html = process_tag($1)
+
+            if html && html != node.content
+              node.replace(html)
+            end
           end
         end
 
@@ -73,6 +82,12 @@ module Banzai
 
       private
 
+      # Replace an entire `[[<em>TOC</em>]]` node with the result generated by
+      # TableOfContentsFilter
+      def process_toc_tag(node)
+        node.parent.parent.replace(result[:toc])
+      end
+
       # Process a single tag into its final HTML form.
       #
       # tag - The String tag contents (the stuff inside the double brackets).
@@ -108,6 +123,12 @@ module Banzai
         end
       end
 
+      def toc_tag?(node)
+        node.content == 'TOC' &&
+          node.parent.name == 'em' &&
+          node.parent.parent.text == '[[TOC]]'
+      end
+
       def image?(path)
         path =~ ALLOWED_IMAGE_EXTENSIONS
       end
diff --git a/lib/banzai/pipeline/wiki_pipeline.rb b/lib/banzai/pipeline/wiki_pipeline.rb
index 50b5450e70b..bc2bf111971 100644
--- a/lib/banzai/pipeline/wiki_pipeline.rb
+++ b/lib/banzai/pipeline/wiki_pipeline.rb
@@ -4,7 +4,11 @@ module Banzai
   module Pipeline
     class WikiPipeline < FullPipeline
       def self.filters
-        super.insert(1, Filter::GollumTagsFilter)
+        @filters ||= begin
+                       filters = super
+                       toc = filters.index(Filter::TableOfContentsFilter)
+                       filters.insert(toc + 1, Filter::GollumTagsFilter)
+                     end
       end
     end
   end
diff --git a/spec/lib/banzai/filter/gollum_tags_filter_spec.rb b/spec/lib/banzai/filter/gollum_tags_filter_spec.rb
index 38baa819957..11cc290d6d0 100644
--- a/spec/lib/banzai/filter/gollum_tags_filter_spec.rb
+++ b/spec/lib/banzai/filter/gollum_tags_filter_spec.rb
@@ -86,4 +86,56 @@ describe Banzai::Filter::GollumTagsFilter, lib: true do
       expect(doc.at_css('a')['href']).to eq 'wiki-slug'
     end
   end
+
+  context 'table of contents' do
+    let(:pipeline) { Banzai::Pipeline[:wiki] }
+
+    it 'replaces the tag with the TableOfContentsFilter result' do
+      markdown = <<-MD.strip_heredoc
+        [[_TOC_]]
+
+        ## Header
+
+        Foo
+      MD
+
+      result = pipeline.call(markdown, project_wiki: project_wiki, project: project)
+
+      aggregate_failures do
+        expect(result[:output].text).not_to include '[[_TOC_]]'
+        expect(result[:output].text).not_to include '[['
+        expect(result[:output].to_html).to include(result[:toc])
+      end
+    end
+
+    it 'is case-sensitive' do
+      markdown = <<-MD.strip_heredoc
+        [[_toc_]]
+
+        # Header 1
+
+        Foo
+      MD
+
+      output = pipeline.to_html(markdown, project_wiki: project_wiki, project: project)
+
+      expect(output).to include('[[<em>toc</em>]]')
+    end
+
+    it 'handles an empty pipeline result' do
+      # No Markdown headers in this doc, so `result[:toc]` will be empty
+      markdown = <<-MD.strip_heredoc
+        [[_TOC_]]
+
+        Foo
+      MD
+
+      output = pipeline.to_html(markdown, project_wiki: project_wiki, project: project)
+
+      aggregate_failures do
+        expect(output).not_to include('<ul>')
+        expect(output).to include('[[<em>TOC</em>]]')
+      end
+    end
+  end
 end