diff --git a/pkg/kubelet/dockertools/manager.go b/pkg/kubelet/dockertools/manager.go
index d71d101fbc3..b36f14d66f6 100644
--- a/pkg/kubelet/dockertools/manager.go
+++ b/pkg/kubelet/dockertools/manager.go
@@ -1724,8 +1724,10 @@ func (dm *DockerManager) computePodContainerChanges(pod *api.Pod, runningPod kub
 			containersToKeep[containerID] = index
 			continue
 		}
-		glog.Infof("pod %q container %q is unhealthy (probe result: %v), it will be killed and re-created.", podFullName, container.Name, result)
-		containersToStart[index] = empty{}
+		if pod.Spec.RestartPolicy != api.RestartPolicyNever {
+			glog.Infof("pod %q container %q is unhealthy (probe result: %v), it will be killed and re-created.", podFullName, container.Name, result)
+			containersToStart[index] = empty{}
+		}
 	}
 
 	// After the loop one of the following should be true:
diff --git a/pkg/kubelet/rkt/rkt.go b/pkg/kubelet/rkt/rkt.go
index 1f9c9a13b13..0ef700bc606 100644
--- a/pkg/kubelet/rkt/rkt.go
+++ b/pkg/kubelet/rkt/rkt.go
@@ -1022,7 +1022,7 @@ func (r *runtime) SyncPod(pod *api.Pod, runningPod kubecontainer.Pod, podStatus
 
 		result, err := r.prober.ProbeLiveness(pod, podStatus, container, string(c.ID), c.Created)
 		// TODO(vmarmol): examine this logic.
-		if err == nil && result != probe.Success {
+		if err == nil && result != probe.Success && pod.Spec.RestartPolicy != api.RestartPolicyNever {
 			glog.Infof("Pod %q container %q is unhealthy (probe result: %v), it will be killed and re-created.", podFullName, container.Name, result)
 			restartPod = true
 			break