diff --git a/frontend_tests/node_tests/unread.js b/frontend_tests/node_tests/unread.js
index 22b5702ef6..def47fe68c 100644
--- a/frontend_tests/node_tests/unread.js
+++ b/frontend_tests/node_tests/unread.js
@@ -261,7 +261,7 @@ run_test('num_unread_for_topic', () => {
     var message = {
         type: 'stream',
         stream_id: stream_id,
-        subject: 'lunch',
+        subject: 'LuncH',
         unread: true,
     };
 
@@ -283,6 +283,26 @@ run_test('num_unread_for_topic', () => {
     msg_ids = unread.get_msg_ids_for_stream(stream_id);
     assert.deepEqual(msg_ids, _.range(1, 501));
 
+    var topic_dict = new Dict({fold_case: true});
+
+    var missing_topics = unread.get_missing_topics({
+        stream_id: stream_id,
+        topic_dict: topic_dict,
+    });
+
+    assert.deepEqual(missing_topics, [
+        { pretty_name: 'LuncH', message_id: 500 },
+    ]);
+
+    topic_dict.set('lUNCh', 'whatever');
+
+    missing_topics = unread.get_missing_topics({
+        stream_id: stream_id,
+        topic_dict: topic_dict,
+    });
+
+    assert.deepEqual(missing_topics, []);
+
     for (i = 0; i < num_msgs; i += 1) {
         message.id = i+1;
         unread.mark_as_read(message.id);
@@ -580,6 +600,13 @@ run_test('empty_cases', () => {
     msg_ids = unread.get_msg_ids_for_stream(stream_id);
     assert.deepEqual(msg_ids, []);
     assert.deepEqual(unread.get_all_msg_ids(), []);
+
+    const missing_topics = unread.get_missing_topics({
+        stream_id: stream_id,
+        topic_dict: 'should-never-be-referenced',
+    });
+    assert.deepEqual(missing_topics, []);
+
 });
 
 run_test('errors', () => {
diff --git a/static/js/unread.js b/static/js/unread.js
index da47f838c3..cb11c956be 100644
--- a/static/js/unread.js
+++ b/static/js/unread.js
@@ -46,6 +46,10 @@ function make_id_set() {
         return ids.keys();
     };
 
+    self.max = function () {
+        return _.max(ids.keys());
+    };
+
     self.is_empty = function () {
         return ids.is_empty();
     };
@@ -100,6 +104,10 @@ function make_bucketer(options) {
         key_to_bucket.each(callback);
     };
 
+    self.keys = function () {
+        return key_to_bucket.keys();
+    };
+
     return self;
 }
 
@@ -304,6 +312,33 @@ exports.unread_topic_counter = (function () {
         return res;
     };
 
+    self.get_missing_topics = function (opts) {
+        var stream_id = opts.stream_id;
+        var topic_dict = opts.topic_dict;
+
+        var per_stream_bucketer = bucketer.get_bucket(stream_id);
+        if (!per_stream_bucketer) {
+            return [];
+        }
+
+        var topic_names = per_stream_bucketer.keys();
+
+        topic_names = _.reject(topic_names, function (topic_name) {
+            return topic_dict.has(topic_name);
+        });
+
+        var result = _.map(topic_names, function (topic_name) {
+            var msgs = per_stream_bucketer.get_bucket(topic_name);
+
+            return {
+                pretty_name: topic_name,
+                message_id: msgs.max(),
+            };
+        });
+
+        return result;
+    };
+
     self.get_stream_count = function (stream_id) {
         var stream_count = 0;
 
@@ -545,6 +580,10 @@ exports.get_all_msg_ids = function () {
     return util.sorted_ids(ids);
 };
 
+exports.get_missing_topics = function (opts) {
+    return exports.unread_topic_counter.get_missing_topics(opts);
+};
+
 exports.get_msg_ids_for_starred = function () {
     // This is here for API consistency sake--we never
     // have unread starred messages.  (Some day we may ironically