diff --git a/virtual-kubelet/commands/root/flag.go b/virtual-kubelet/commands/root/flag.go
index 3a5bc74..e599fd9 100644
--- a/virtual-kubelet/commands/root/flag.go
+++ b/virtual-kubelet/commands/root/flag.go
@@ -78,7 +78,7 @@ func installFlags(flags *pflag.FlagSet, c *Opts) {
 	flags.StringVar(&c.TraceSampleRate, "trace-sample-rate", c.TraceSampleRate, "set probability of tracing samples")
 
 	flags.DurationVar(&c.InformerResyncPeriod, "full-resync-period", c.InformerResyncPeriod, "how often to perform a full resync of pods between kubernetes and the provider")
-
+	flags.DurationVar(&c.StartupTimeout, "startup-timeout", c.StartupTimeout, "How long to wait for the virtual-kubelet to start")
 }
 
 func getEnv(key, defaultValue string) string {
diff --git a/virtual-kubelet/commands/root/opts.go b/virtual-kubelet/commands/root/opts.go
index 41038fa..845f404 100644
--- a/virtual-kubelet/commands/root/opts.go
+++ b/virtual-kubelet/commands/root/opts.go
@@ -78,6 +78,9 @@ type Opts struct {
 	TraceExporters  []string
 	TraceSampleRate string
 	TraceConfig     opencensus.TracingExporterOptions
+
+	// Startup Timeout is how long to wait for the kubelet to start
+	StartupTimeout time.Duration
 }
 
 // SetDefaultOpts sets default options for unset values on the passed in option struct.
diff --git a/virtual-kubelet/commands/root/root.go b/virtual-kubelet/commands/root/root.go
index 099e81f..caacdf6 100644
--- a/virtual-kubelet/commands/root/root.go
+++ b/virtual-kubelet/commands/root/root.go
@@ -17,6 +17,7 @@ package root
 import (
 	"context"
 	"os"
+	"time"
 
 	"github.com/cpuguy83/strongerrors"
 	"github.com/pkg/errors"
@@ -54,6 +55,9 @@ This allows users to schedule kubernetes workloads on nodes that aren't running
 }
 
 func runRootCommand(ctx context.Context, c Opts) error {
+	ctx, cancel := context.WithCancel(ctx)
+	defer cancel()
+
 	if ok := providers.ValidOperatingSystems[c.OperatingSystem]; !ok {
 		return strongerrors.InvalidArgument(errors.Errorf("operating system %q is not supported", c.OperatingSystem))
 	}
@@ -166,6 +170,16 @@ func runRootCommand(ctx context.Context, c Opts) error {
 		}
 	}()
 
+	if c.StartupTimeout > 0 {
+		// If there is a startup timeout, it does two things:
+		// 1. It causes the VK to shutdown if we haven't gotten into an operational state in a time period
+		// 2. It prevents node advertisement from happening until we're in an operational state
+		err = waitForVK(ctx, c.StartupTimeout, vk)
+		if err != nil {
+			return err
+		}
+	}
+
 	go func() {
 		if err := node.Run(ctx); err != nil {
 			log.G(ctx).Fatal(err)
@@ -178,6 +192,21 @@ func runRootCommand(ctx context.Context, c Opts) error {
 	return nil
 }
 
+func waitForVK(ctx context.Context, time time.Duration, vk *vkubelet.Server) error {
+	ctx, cancel := context.WithTimeout(ctx, time)
+	defer cancel()
+
+	// Wait for the VK / PC close the the ready channel, or time out and return
+	log.G(ctx).Info("Waiting for pod controller / VK to be ready")
+
+	select {
+	case <-vk.Ready():
+		return nil
+	case <-ctx.Done():
+		return errors.Wrap(ctx.Err(), "Error while starting up VK")
+	}
+}
+
 func newClient(configPath string) (*kubernetes.Clientset, error) {
 	var config *rest.Config