diff --git a/dataproc/create_cluster_and_submit_job.py b/dataproc/create_cluster_and_submit_job.py
index 0cf58ad74c53..377baa6452c0 100644
--- a/dataproc/create_cluster_and_submit_job.py
+++ b/dataproc/create_cluster_and_submit_job.py
@@ -119,6 +119,35 @@ def get_cluster_id_by_name(cluster_list, cluster_name):
     cluster = [c for c in cluster_list if c['clusterName'] == cluster_name][0]
     return cluster['clusterUuid'], cluster['config']['configBucket']
 
+# [START update_cluster]
+def update_cluster(dataproc, project, cluster_name, zone, num_workers):
+    """Updates the specified cluster to the given number of workers.
+    Use 'config.worker_config.num_instances' and 'workerConfig' for workers.
+    Use 'config.secondary_worker_config.num_instances' and
+    'secondaryWorkerConfig' for preemptible workers."""
+    print('Updating cluster to %d workers.' % num_workers)
+    zone_uri = \
+        'https://www.googleapis.com/compute/v1/projects/{}/zones/{}'.format(
+            project, zone)
+    cluster_data = {
+        'projectId': project,
+        'clusterName': cluster_name,
+        'config': {
+            'gceClusterConfig': {
+                'zoneUri': zone_uri
+            },
+            'workerConfig': {
+                'numInstances': num_workers
+            }
+        }
+    }
+    result = dataproc.projects().regions().clusters().patch(
+        projectId = project,
+        region = REGION,
+        clusterName = cluster_name,
+        updateMask = 'config.worker_config.num_instances',
+        body = cluster_data).execute()
+# [END update_cluster]
 
 # [START submit_pyspark_job]
 def submit_pyspark_job(dataproc, project, cluster_name, bucket_name, filename):