diff --git a/pkg/controller/deployment/deployment_controller.go b/pkg/controller/deployment/deployment_controller.go
index d9b996c2ee8..5d0d777b058 100644
--- a/pkg/controller/deployment/deployment_controller.go
+++ b/pkg/controller/deployment/deployment_controller.go
@@ -886,29 +886,32 @@ func (dc *DeploymentController) reconcileOldReplicaSets(allRSs []*extensions.Rep
 
 	// Clean up unhealthy replicas first, otherwise unhealthy replicas will block deployment
 	// and cause timeout. See https://github.com/kubernetes/kubernetes/issues/16737
-	cleanupCount, err := dc.cleanupUnhealthyReplicas(oldRSs, deployment, maxScaledDown)
+	oldRSs, cleanupCount, err := dc.cleanupUnhealthyReplicas(oldRSs, deployment, maxScaledDown)
 	if err != nil {
 		return false, nil
 	}
+	glog.V(4).Infof("Cleaned up unhealthy replicas from old RSes by %d", cleanupCount)
 
 	// Scale down old replica sets, need check maxUnavailable to ensure we can scale down
+	allRSs = append(oldRSs, newRS)
 	scaledDownCount, err := dc.scaleDownOldReplicaSetsForRollingUpdate(allRSs, oldRSs, deployment)
 	if err != nil {
 		return false, nil
 	}
+	glog.V(4).Infof("Scaled down old RSes by %d", scaledDownCount)
 
 	totalScaledDown := cleanupCount + scaledDownCount
 	return totalScaledDown > 0, nil
 }
 
 // cleanupUnhealthyReplicas will scale down old replica sets with unhealthy replicas, so that all unhealthy replicas will be deleted.
-func (dc *DeploymentController) cleanupUnhealthyReplicas(oldRSs []*extensions.ReplicaSet, deployment *extensions.Deployment, maxCleanupCount int) (int, error) {
+func (dc *DeploymentController) cleanupUnhealthyReplicas(oldRSs []*extensions.ReplicaSet, deployment *extensions.Deployment, maxCleanupCount int) ([]*extensions.ReplicaSet, int, error) {
 	sort.Sort(controller.ReplicaSetsByCreationTimestamp(oldRSs))
 	// Safely scale down all old replica sets with unhealthy replicas. Replica set will sort the pods in the order
 	// such that not-ready < ready, unscheduled < scheduled, and pending < running. This ensures that unhealthy replicas will
 	// been deleted first and won't increase unavailability.
 	totalScaledDown := 0
-	for _, targetRS := range oldRSs {
+	for i, targetRS := range oldRSs {
 		if totalScaledDown >= maxCleanupCount {
 			break
 		}
@@ -918,7 +921,7 @@ func (dc *DeploymentController) cleanupUnhealthyReplicas(oldRSs []*extensions.Re
 		}
 		readyPodCount, err := deploymentutil.GetAvailablePodsForReplicaSets(dc.client, []*extensions.ReplicaSet{targetRS}, 0)
 		if err != nil {
-			return totalScaledDown, fmt.Errorf("could not find available pods: %v", err)
+			return nil, totalScaledDown, fmt.Errorf("could not find available pods: %v", err)
 		}
 		if targetRS.Spec.Replicas == readyPodCount {
 			// no unhealthy replicas found, no scaling required.
@@ -927,13 +930,17 @@ func (dc *DeploymentController) cleanupUnhealthyReplicas(oldRSs []*extensions.Re
 
 		scaledDownCount := integer.IntMin(maxCleanupCount-totalScaledDown, targetRS.Spec.Replicas-readyPodCount)
 		newReplicasCount := targetRS.Spec.Replicas - scaledDownCount
-		_, _, err = dc.scaleReplicaSetAndRecordEvent(targetRS, newReplicasCount, deployment)
+		if newReplicasCount > targetRS.Spec.Replicas {
+			return nil, 0, fmt.Errorf("when cleaning up unhealthy replicas, got invalid request to scale down %s/%s %d -> %d", targetRS.Namespace, targetRS.Name, targetRS.Spec.Replicas, newReplicasCount)
+		}
+		_, updatedOldRS, err := dc.scaleReplicaSetAndRecordEvent(targetRS, newReplicasCount, deployment)
 		if err != nil {
-			return totalScaledDown, err
+			return nil, totalScaledDown, err
 		}
 		totalScaledDown += scaledDownCount
+		oldRSs[i] = updatedOldRS
 	}
-	return totalScaledDown, nil
+	return oldRSs, totalScaledDown, nil
 }
 
 // scaleDownOldReplicaSetsForRollingUpdate scales down old replica sets when deployment strategy is "RollingUpdate".
@@ -973,6 +980,9 @@ func (dc *DeploymentController) scaleDownOldReplicaSetsForRollingUpdate(allRSs [
 		// Scale down.
 		scaleDownCount := integer.IntMin(targetRS.Spec.Replicas, totalScaleDownCount-totalScaledDown)
 		newReplicasCount := targetRS.Spec.Replicas - scaleDownCount
+		if newReplicasCount > targetRS.Spec.Replicas {
+			return 0, fmt.Errorf("when scaling down old RS, got invalid request to scale down %s/%s %d -> %d", targetRS.Namespace, targetRS.Name, targetRS.Spec.Replicas, newReplicasCount)
+		}
 		_, _, err = dc.scaleReplicaSetAndRecordEvent(targetRS, newReplicasCount, deployment)
 		if err != nil {
 			return totalScaledDown, err
diff --git a/pkg/controller/deployment/deployment_controller_test.go b/pkg/controller/deployment/deployment_controller_test.go
index 31cb70061c4..648e7e633e6 100644
--- a/pkg/controller/deployment/deployment_controller_test.go
+++ b/pkg/controller/deployment/deployment_controller_test.go
@@ -471,7 +471,7 @@ func TestDeploymentController_cleanupUnhealthyReplicas(t *testing.T) {
 			client:        &fakeClientset,
 			eventRecorder: &record.FakeRecorder{},
 		}
-		cleanupCount, err := controller.cleanupUnhealthyReplicas(oldRSs, &deployment, test.maxCleanupCount)
+		_, cleanupCount, err := controller.cleanupUnhealthyReplicas(oldRSs, &deployment, test.maxCleanupCount)
 		if err != nil {
 			t.Errorf("unexpected error: %v", err)
 			continue
@@ -522,6 +522,13 @@ func TestDeploymentController_scaleDownOldReplicaSetsForRollingUpdate(t *testing
 			oldReplicas:        0,
 			scaleExpected:      false,
 		},
+		{
+			deploymentReplicas: 10,
+			maxUnavailable:     intstr.FromInt(2),
+			readyPods:          1,
+			oldReplicas:        10,
+			scaleExpected:      false,
+		},
 	}
 
 	for i, test := range tests {
diff --git a/test/e2e/deployment.go b/test/e2e/deployment.go
index 897813f445f..975392bb1a9 100644
--- a/test/e2e/deployment.go
+++ b/test/e2e/deployment.go
@@ -182,7 +182,7 @@ func stopDeployment(c *clientset.Clientset, oldC client.Interface, ns, deploymen
 	_, err = c.Extensions().Deployments(ns).Get(deployment.Name)
 	Expect(err).To(HaveOccurred())
 	Expect(errors.IsNotFound(err)).To(BeTrue())
-	Logf("ensuring deployment %s rcs were deleted", deploymentName)
+	Logf("ensuring deployment %s RSes were deleted", deploymentName)
 	selector, err := unversioned.LabelSelectorAsSelector(deployment.Spec.Selector)
 	Expect(err).NotTo(HaveOccurred())
 	options := api.ListOptions{LabelSelector: selector}