diff --git a/test/e2e/testing-manifests/statefulset/etcd/pdb.yaml b/test/e2e/testing-manifests/statefulset/etcd/pdb.yaml
new file mode 100644
index 00000000000..64d68aac7c0
--- /dev/null
+++ b/test/e2e/testing-manifests/statefulset/etcd/pdb.yaml
@@ -0,0 +1,11 @@
+apiVersion: policy/v1beta1
+kind: PodDisruptionBudget
+metadata:
+  name: etcd-pdb
+  labels:
+    pdb: etcd
+spec:
+  minAvailable: 2
+  selector:
+    matchLabels:
+      app: etcd
diff --git a/test/e2e/testing-manifests/statefulset/etcd/service.yaml b/test/e2e/testing-manifests/statefulset/etcd/service.yaml
new file mode 100644
index 00000000000..3932f04f42b
--- /dev/null
+++ b/test/e2e/testing-manifests/statefulset/etcd/service.yaml
@@ -0,0 +1,18 @@
+apiVersion: v1
+kind: Service
+metadata:
+  annotations:
+    service.alpha.kubernetes.io/tolerate-unready-endpoints: "true"
+metadata:
+  name: etcd
+  labels:
+    app: etcd
+spec:
+  ports:
+  - port: 2380
+    name: etcd-server
+  - port: 2379
+    name: etcd-client
+  clusterIP: None
+  selector:
+    app: etcd
diff --git a/test/e2e/testing-manifests/statefulset/etcd/statefulset.yaml b/test/e2e/testing-manifests/statefulset/etcd/statefulset.yaml
new file mode 100644
index 00000000000..c7800cd4e8f
--- /dev/null
+++ b/test/e2e/testing-manifests/statefulset/etcd/statefulset.yaml
@@ -0,0 +1,178 @@
+apiVersion: apps/v1beta1
+kind: StatefulSet
+metadata:
+  name: etcd
+  labels:
+    app: etcd
+spec:
+  serviceName: etcd
+  replicas: 3
+  template:
+    metadata:
+      name: etcd
+      labels:
+        app: etcd
+    spec:
+      containers:
+      - name: etcd
+        image: gcr.io/google_containers/etcd-amd64:2.2.5
+        imagePullPolicy: Always
+        ports:
+        - containerPort: 2380
+          name: peer
+        - containerPort: 2379
+          name: client
+        resources:
+          requests:
+            cpu: 100m
+            memory: 512Mi
+        env:
+        - name: INITIAL_CLUSTER_SIZE
+          value: "3"
+        - name: SET_NAME
+          value: etcd
+        volumeMounts:
+        - name: datadir
+          mountPath: /var/run/etcd
+        lifecycle:
+          preStop:
+            exec:
+              command:
+                - "/bin/sh"
+                - "-ec"
+                - |
+                  EPS=""
+                  for i in $(seq 0 $((${INITIAL_CLUSTER_SIZE} - 1))); do
+                      EPS="${EPS}${EPS:+,}http://${SET_NAME}-${i}.${SET_NAME}:2379"
+                  done
+
+                  HOSTNAME=$(hostname)
+
+                  member_hash() {
+                      etcdctl member list | grep http://${HOSTNAME}.${SET_NAME}:2380 | cut -d':' -f1 | cut -d'[' -f1
+                  }
+
+                  echo "Removing ${HOSTNAME} from etcd cluster"
+
+                  ETCDCTL_ENDPOINT=${EPS} etcdctl member remove $(member_hash)
+                  if [ $? -eq 0 ]; then
+                      # Remove everything otherwise the cluster will no longer scale-up
+                      rm -rf /var/run/etcd/*
+                  fi
+        command:
+          - "/bin/sh"
+          - "-ec"
+          - |
+            HOSTNAME=$(hostname)
+
+            # store member id into PVC for later member replacement
+            collect_member() {
+                while ! etcdctl member list &>/dev/null; do sleep 1; done
+                etcdctl member list | grep http://${HOSTNAME}.${SET_NAME}:2380 | cut -d':' -f1 | cut -d'[' -f1 > /var/run/etcd/member_id
+                exit 0
+            }
+
+            eps() {
+                EPS=""
+                for i in $(seq 0 $((${INITIAL_CLUSTER_SIZE} - 1))); do
+                    EPS="${EPS}${EPS:+,}http://${SET_NAME}-${i}.${SET_NAME}:2379"
+                done
+                echo ${EPS}
+            }
+
+            member_hash() {
+                etcdctl member list | grep http://${HOSTNAME}.${SET_NAME}:2380 | cut -d':' -f1 | cut -d'[' -f1
+            }
+
+            # re-joining after failure?
+            if [ -e /var/run/etcd/default.etcd ]; then
+                echo "Re-joining etcd member"
+                member_id=$(cat /var/run/etcd/member_id)
+
+                # re-join member
+                ETCDCTL_ENDPOINT=$(eps) etcdctl member update ${member_id} http://${HOSTNAME}.${SET_NAME}:2380
+                exec etcd --name ${HOSTNAME} \
+                    --listen-peer-urls http://${HOSTNAME}.${SET_NAME}:2380 \
+                    --listen-client-urls http://${HOSTNAME}.${SET_NAME}:2379,http://127.0.0.1:2379 \
+                    --advertise-client-urls http://${HOSTNAME}.${SET_NAME}:2379 \
+                    --data-dir /var/run/etcd/default.etcd
+            fi
+
+            # etcd-SET_ID
+            SET_ID=${HOSTNAME:5:${#HOSTNAME}}
+
+            # adding a new member to existing cluster (assuming all initial pods are available)
+            if [ "${SET_ID}" -ge ${INITIAL_CLUSTER_SIZE} ]; then
+                export ETCDCTL_ENDPOINT=$(eps)
+
+                # member already added?
+                MEMBER_HASH=$(member_hash)
+                if [ -n "${MEMBER_HASH}" ]; then
+                    # the member hash exists but for some reason etcd failed
+                    # as the datadir has not be created, we can remove the member
+                    # and retrieve new hash
+                    etcdctl member remove ${MEMBER_HASH}
+                fi
+
+                echo "Adding new member"
+                etcdctl member add ${HOSTNAME} http://${HOSTNAME}.${SET_NAME}:2380 | grep "^ETCD_" > /var/run/etcd/new_member_envs
+
+                if [ $? -ne 0 ]; then
+                    echo "Exiting"
+                    rm -f /var/run/etcd/new_member_envs
+                    exit 1
+                fi
+
+                cat /var/run/etcd/new_member_envs
+                source /var/run/etcd/new_member_envs
+
+                collect_member &
+
+                exec etcd --name ${HOSTNAME} \
+                    --listen-peer-urls http://${HOSTNAME}.${SET_NAME}:2380 \
+                    --listen-client-urls http://${HOSTNAME}.${SET_NAME}:2379,http://127.0.0.1:2379 \
+                    --advertise-client-urls http://${HOSTNAME}.${SET_NAME}:2379 \
+                    --data-dir /var/run/etcd/default.etcd \
+                    --initial-advertise-peer-urls http://${HOSTNAME}.${SET_NAME}:2380 \
+                    --initial-cluster ${ETCD_INITIAL_CLUSTER} \
+                    --initial-cluster-state ${ETCD_INITIAL_CLUSTER_STATE}
+            fi
+
+            for i in $(seq 0 $((${INITIAL_CLUSTER_SIZE} - 1))); do
+                while true; do
+                    echo "Waiting for ${SET_NAME}-${i}.${SET_NAME} to come up"
+                    ping -W 1 -c 1 ${SET_NAME}-${i}.${SET_NAME} > /dev/null && break
+                    sleep 1s
+                done
+            done
+
+            PEERS=""
+            for i in $(seq 0 $((${INITIAL_CLUSTER_SIZE} - 1))); do
+                PEERS="${PEERS}${PEERS:+,}${SET_NAME}-${i}=http://${SET_NAME}-${i}.${SET_NAME}:2380"
+            done
+
+            collect_member &
+
+            # join member
+            exec etcd --name ${HOSTNAME} \
+                --initial-advertise-peer-urls http://${HOSTNAME}.${SET_NAME}:2380 \
+                --listen-peer-urls http://${HOSTNAME}.${SET_NAME}:2380 \
+                --listen-client-urls http://${HOSTNAME}.${SET_NAME}:2379,http://127.0.0.1:2379 \
+                --advertise-client-urls http://${HOSTNAME}.${SET_NAME}:2379 \
+                --initial-cluster-token etcd-cluster-1 \
+                --initial-cluster ${PEERS} \
+                --initial-cluster-state new \
+                --data-dir /var/run/etcd/default.etcd
+  volumeClaimTemplates:
+  - metadata:
+      name: datadir
+      annotations:
+        volume.alpha.kubernetes.io/storage-class: anything
+    spec:
+      accessModes:
+        - "ReadWriteOnce"
+      resources:
+        requests:
+          # upstream recommended max is 700M
+          storage: 1Gi
+
diff --git a/test/e2e/testing-manifests/statefulset/etcd/tester.yaml b/test/e2e/testing-manifests/statefulset/etcd/tester.yaml
new file mode 100644
index 00000000000..c5ea0b90c14
--- /dev/null
+++ b/test/e2e/testing-manifests/statefulset/etcd/tester.yaml
@@ -0,0 +1,24 @@
+apiVersion: apps/v1beta1
+kind: Deployment
+metadata:
+  name: etcd-test-server
+spec:
+  replicas: 3
+  template:
+    metadata:
+      labels:
+        app: test-server
+    spec:
+      containers:
+      - name: test-server
+        image: gcr.io/google-containers/etcd-statefulset-e2e-test:0.0
+        imagePullPolicy: Always
+        ports:
+        - containerPort: 8080
+        readinessProbe:
+          httpGet:
+            path: /healthz
+            port: 8080
+          initialDelaySeconds: 2
+          periodSeconds: 2
+
diff --git a/test/e2e/upgrades/BUILD b/test/e2e/upgrades/BUILD
index 1b0cc573751..98082c71864 100644
--- a/test/e2e/upgrades/BUILD
+++ b/test/e2e/upgrades/BUILD
@@ -14,6 +14,7 @@ go_library(
         "configmaps.go",
         "daemonsets.go",
         "deployments.go",
+        "etcd.go",
         "horizontal_pod_autoscalers.go",
         "ingress.go",
         "job.go",
diff --git a/test/e2e/upgrades/etcd.go b/test/e2e/upgrades/etcd.go
new file mode 100644
index 00000000000..aea9e08ec64
--- /dev/null
+++ b/test/e2e/upgrades/etcd.go
@@ -0,0 +1,199 @@
+/*
+Copyright 2017 The Kubernetes Authors.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+*/
+
+package upgrades
+
+import (
+	"encoding/json"
+	"fmt"
+	"io/ioutil"
+	"net/http"
+	"path/filepath"
+	"sync"
+	"time"
+
+	. "github.com/onsi/ginkgo"
+	. "github.com/onsi/gomega"
+
+	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
+	"k8s.io/apimachinery/pkg/util/wait"
+	"k8s.io/kubernetes/pkg/util/version"
+	"k8s.io/kubernetes/test/e2e/framework"
+)
+
+const manifestPath = "test/e2e/testing-manifests/statefulset/etcd"
+
+type EtcdUpgradeTest struct {
+	ip               string
+	successfulWrites int
+	ssTester         *framework.StatefulSetTester
+}
+
+func (EtcdUpgradeTest) Name() string { return "etcd-upgrade" }
+
+func (EtcdUpgradeTest) Skip(upgCtx UpgradeContext) bool {
+	minVersion := version.MustParseSemantic("1.6.0")
+	for _, vCtx := range upgCtx.Versions {
+		if vCtx.Version.LessThan(minVersion) {
+			return true
+		}
+	}
+	return false
+}
+
+func kubectlCreate(ns, file string) {
+	path := filepath.Join(framework.TestContext.RepoRoot, manifestPath, file)
+	framework.RunKubectlOrDie("create", "-f", path, fmt.Sprintf("--namespace=%s", ns))
+}
+
+func (t *EtcdUpgradeTest) Setup(f *framework.Framework) {
+	ns := f.Namespace.Name
+	statefulsetPoll := 30 * time.Second
+	statefulsetTimeout := 10 * time.Minute
+	t.ssTester = framework.NewStatefulSetTester(f.ClientSet)
+
+	By("Creating a PDB")
+	kubectlCreate(ns, "pdb.yaml")
+
+	By("Creating an etcd StatefulSet")
+	t.ssTester.CreateStatefulSet(manifestPath, ns)
+
+	By("Creating an etcd--test-server deployment")
+	kubectlCreate(ns, "tester.yaml")
+
+	By("Getting the ingress IPs from the services")
+	err := wait.PollImmediate(statefulsetPoll, statefulsetTimeout, func() (bool, error) {
+		if t.ip = t.getServiceIP(f, ns, "test-server"); t.ip == "" {
+			return false, nil
+		}
+		if _, err := t.listUsers(); err != nil {
+			framework.Logf("Service endpoint is up but isn't responding")
+			return false, nil
+		}
+		return true, nil
+	})
+	Expect(err).NotTo(HaveOccurred())
+	framework.Logf("Service endpoint is up")
+
+	By("Adding 2 dummy users")
+	Expect(t.addUser("Alice")).NotTo(HaveOccurred())
+	Expect(t.addUser("Bob")).NotTo(HaveOccurred())
+	t.successfulWrites = 2
+
+	By("Verifying that the users exist")
+	users, err := t.listUsers()
+	Expect(err).NotTo(HaveOccurred())
+	Expect(len(users)).To(Equal(2))
+}
+
+func (t *EtcdUpgradeTest) listUsers() ([]string, error) {
+	r, err := http.Get(fmt.Sprintf("http://%s:8080/list", t.ip))
+	if err != nil {
+		return nil, err
+	}
+	defer r.Body.Close()
+	if r.StatusCode != http.StatusOK {
+		b, err := ioutil.ReadAll(r.Body)
+		if err != nil {
+			return nil, err
+		}
+		return nil, fmt.Errorf(string(b))
+	}
+	var names []string
+	if err := json.NewDecoder(r.Body).Decode(&names); err != nil {
+		return nil, err
+	}
+	return names, nil
+}
+
+func (t *EtcdUpgradeTest) addUser(name string) error {
+	val := map[string][]string{"name": {name}}
+	r, err := http.PostForm(fmt.Sprintf("http://%s:8080/add", t.ip), val)
+	if err != nil {
+		return err
+	}
+	defer r.Body.Close()
+	if r.StatusCode != http.StatusOK {
+		b, err := ioutil.ReadAll(r.Body)
+		if err != nil {
+			return err
+		}
+		return fmt.Errorf(string(b))
+	}
+	return nil
+}
+
+func (t *EtcdUpgradeTest) getServiceIP(f *framework.Framework, ns, svcName string) string {
+	svc, err := f.ClientSet.CoreV1().Services(ns).Get(svcName, metav1.GetOptions{})
+	Expect(err).NotTo(HaveOccurred())
+	ingress := svc.Status.LoadBalancer.Ingress
+	if len(ingress) == 0 {
+		return ""
+	}
+	return ingress[0].IP
+}
+
+func (t *EtcdUpgradeTest) Test(f *framework.Framework, done <-chan struct{}, upgrade UpgradeType) {
+	By("Continuously polling the database during upgrade.")
+	var (
+		success, failures, writeAttempts, lastUserCount int
+		mu                                              sync.Mutex
+		errors                                          = map[string]int{}
+	)
+	// Write loop.
+	go wait.Until(func() {
+		writeAttempts++
+		if err := t.addUser(fmt.Sprintf("user-%d", writeAttempts)); err != nil {
+			framework.Logf("Unable to add user: %v", err)
+			mu.Lock()
+			errors[err.Error()]++
+			mu.Unlock()
+			return
+		}
+		t.successfulWrites++
+	}, 10*time.Millisecond, done)
+	// Read loop.
+	wait.Until(func() {
+		users, err := t.listUsers()
+		if err != nil {
+			framework.Logf("Could not retrieve users: %v", err)
+			failures++
+			mu.Lock()
+			errors[err.Error()]++
+			mu.Unlock()
+			return
+		}
+		success++
+		lastUserCount = len(users)
+	}, 10*time.Millisecond, done)
+	framework.Logf("got %d users; want >=%d", lastUserCount, t.successfulWrites)
+
+	Expect(lastUserCount >= t.successfulWrites).To(BeTrue())
+	ratio := float64(success) / float64(success+failures)
+	framework.Logf("Successful gets %d/%d=%v", success, success+failures, ratio)
+	ratio = float64(t.successfulWrites) / float64(writeAttempts)
+	framework.Logf("Successful writes %d/%d=%v", t.successfulWrites, writeAttempts, ratio)
+	framework.Logf("Errors: %v", errors)
+	// TODO(maisem): tweak this value once we have a few test runs.
+	Expect(ratio > 0.75).To(BeTrue())
+}
+
+// Teardown does one final check of the data's availability.
+func (t *EtcdUpgradeTest) Teardown(f *framework.Framework) {
+	users, err := t.listUsers()
+	Expect(err).NotTo(HaveOccurred())
+	Expect(len(users) >= t.successfulWrites).To(BeTrue())
+}