diff --git a/cmd/kube-controller-manager/controller-manager.go b/cmd/kube-controller-manager/controller-manager.go
index 97299a0189c..fd7d52599fb 100644
--- a/cmd/kube-controller-manager/controller-manager.go
+++ b/cmd/kube-controller-manager/controller-manager.go
@@ -49,7 +49,10 @@ var (
 	cloudProvider   = flag.String("cloud_provider", "", "The provider for cloud services.  Empty string for no provider.")
 	cloudConfigFile = flag.String("cloud_config", "", "The path to the cloud provider configuration file.  Empty string for no configuration file.")
 	minionRegexp    = flag.String("minion_regexp", "", "If non empty, and -cloud_provider is specified, a regular expression for matching minion VMs.")
-	machineList     util.StringList
+	nodeSyncPeriod  = flag.Duration("node_sync_period", 10*time.Second, ""+
+		"The period for syncing nodes from cloudprovider. Longer periods will result in "+
+		"fewer calls to cloud provider, but may delay addition of new nodes to cluster.")
+	machineList util.StringList
 	// TODO: Discover these by pinging the host machines, and rip out these flags.
 	// TODO: in the meantime, use resource.QuantityFlag() instead of these
 	nodeMilliCPU = flag.Int64("node_milli_cpu", 1000, "The amount of MilliCPU provisioned on each node")
@@ -107,7 +110,7 @@ func main() {
 		},
 	}
 	nodeController := nodeControllerPkg.NewNodeController(cloud, *minionRegexp, machineList, nodeResources, kubeClient)
-	nodeController.Run(10 * time.Second)
+	nodeController.Run(*nodeSyncPeriod)
 
 	select {}
 }
diff --git a/docs/man/kube-controller-manager.1.md b/docs/man/kube-controller-manager.1.md
index 1e68503ac70..52a2d9ec564 100644
--- a/docs/man/kube-controller-manager.1.md
+++ b/docs/man/kube-controller-manager.1.md
@@ -54,11 +54,14 @@ The kube-controller-manager has several options.
 	log to standard error instead of files.
 
 **--machines**=[]
-        List of machines to schedule onto, comma separated.
+    List of machines to schedule onto, comma separated.
 
 **--master**=""
 	The address of the Kubernetes API server.
 
+**--node_sync_peroid**=10s
+    The period for syncing nodes from cloudprovider.
+
 **--port**=10252
 	The port that the controller-manager's http service runs on.
 
diff --git a/docs/node.md b/docs/node.md
index 8ee566c02fd..c3b6ba068d1 100644
--- a/docs/node.md
+++ b/docs/node.md
@@ -58,7 +58,8 @@ objects. It performs two major functions: cluster-wide node synchronization
 and single node life-cycle management.
 
 Node controller has a sync loop that creates/deletes `Node`s from Kubernetes
-based on all matching VM instances listed from cloud provider. If a new instance
+based on all matching VM instances listed from cloud provider. The sync period
+can be controlled via flag "--node_sync_period". If a new instance
 gets created, Node Controller creates a representation for it. If an existing
 instance gets deleted, Node Controller deletes the representation. Note however,
 Node Controller is unable to provision the node for you, i.e. it won't install