diff --git a/frontend_tests/node_tests/stream_topic_history.js b/frontend_tests/node_tests/stream_topic_history.js
index 8cadbb2c16..b9f8a8c1dc 100644
--- a/frontend_tests/node_tests/stream_topic_history.js
+++ b/frontend_tests/node_tests/stream_topic_history.js
@@ -256,10 +256,14 @@ test("test_unread_logic", () => {
     assert.deepEqual(history, ["toPic1", "topic2"]);
 
     const msgs = [
-        {id: 150, topic: "TOPIC2"}, // will be ignored
+        // This will be ignored as a case variant of `topic2` above.
+        {id: 150, topic: "TOPIC2"},
         {id: 61, topic: "unread1"},
         {id: 60, topic: "unread1"},
         {id: 20, topic: "UNREAD2"},
+        // We're going to mark this as read; this will verify the logic
+        // in unreads.js for only including topics with nonzero unreads.
+        {id: 79, topic: "to_mark_as_read"},
     ];
 
     for (const msg of msgs) {
@@ -269,6 +273,7 @@ test("test_unread_logic", () => {
     }
 
     unread.process_loaded_messages(msgs);
+    unread.mark_as_read(79);
 
     history = stream_topic_history.get_recent_topic_names(stream_id);
     assert.deepEqual(history, ["toPic1", "unread1", "topic2", "UNREAD2"]);
diff --git a/static/js/stream_topic_history.js b/static/js/stream_topic_history.js
index 715c50160c..03a4987310 100644
--- a/static/js/stream_topic_history.js
+++ b/static/js/stream_topic_history.js
@@ -201,6 +201,8 @@ export class PerStreamHistory {
     get_recent_topic_names() {
         const my_recents = Array.from(this.topics.values());
 
+        /* Add any older topics with unreads that may not be present
+         * in our local cache. */
         const missing_topics = unread.get_missing_topics({
             stream_id: this.stream_id,
             topic_dict: this.topics,
diff --git a/static/js/unread.js b/static/js/unread.js
index c218641ec3..08a98f5ba5 100644
--- a/static/js/unread.js
+++ b/static/js/unread.js
@@ -265,6 +265,15 @@ class UnreadTopicCounter {
     }
 
     get_missing_topics(opts) {
+        /* Clients have essentially complete unread data, but
+         * stream_topic_history.is_complete_for_stream_id() can be
+         * false. In that situation, this function helps ensure that
+         * we include all topics with unread messages in data that.
+         *
+         * It will return all topics in the provided stream with a
+         * nonzero unread count that are not already present in the
+         * topic_dict parameter.
+         */
         const stream_id = opts.stream_id;
         const topic_dict = opts.topic_dict;
 
@@ -275,6 +284,13 @@ class UnreadTopicCounter {
 
         let topic_names = Array.from(per_stream_bucketer.keys());
 
+        /* Include topics that have at least one unread. It would likely
+         * be better design for buckets to be deleted when emptied. */
+        topic_names = topic_names.filter((topic_name) => {
+            const messages = Array.from(per_stream_bucketer.get_bucket(topic_name));
+            return messages.length > 0;
+        });
+        /* And aren't already present in topic_dict. */
         topic_names = topic_names.filter((topic_name) => !topic_dict.has(topic_name));
 
         const result = topic_names.map((topic_name) => {