diff --git a/plugin/pkg/scheduler/algorithm/predicates/predicates.go b/plugin/pkg/scheduler/algorithm/predicates/predicates.go
index 31eb3a5449..eff7cc5f4a 100644
--- a/plugin/pkg/scheduler/algorithm/predicates/predicates.go
+++ b/plugin/pkg/scheduler/algorithm/predicates/predicates.go
@@ -20,6 +20,7 @@ import (
 	"fmt"
 	"math/rand"
 	"strconv"
+	"strings"
 	"sync"
 	"time"
 
@@ -109,7 +110,7 @@ type predicateMetadata struct {
 
 func isVolumeConflict(volume v1.Volume, pod *v1.Pod) bool {
 	// fast path if there is no conflict checking targets.
-	if volume.GCEPersistentDisk == nil && volume.AWSElasticBlockStore == nil && volume.RBD == nil {
+	if volume.GCEPersistentDisk == nil && volume.AWSElasticBlockStore == nil && volume.RBD == nil && volume.ISCSI == nil {
 		return false
 	}
 
@@ -128,6 +129,26 @@ func isVolumeConflict(volume v1.Volume, pod *v1.Pod) bool {
 			}
 		}
 
+		if volume.ISCSI != nil && existingVolume.ISCSI != nil {
+			iqn, lun, target := volume.ISCSI.IQN, volume.ISCSI.Lun, volume.ISCSI.TargetPortal
+			eiqn, elun, etarget := existingVolume.ISCSI.IQN, existingVolume.ISCSI.Lun, existingVolume.ISCSI.TargetPortal
+			if !strings.Contains(target, ":") {
+				target = target + ":3260"
+			}
+			if !strings.Contains(etarget, ":") {
+				etarget = etarget + ":3260"
+			}
+			lun1 := strconv.Itoa(int(lun))
+			elun1 := strconv.Itoa(int(elun))
+
+			// two ISCSI volumes are same, if they share the same iqn, lun and target. As iscsi volumes are of type
+			// RWO or ROX, we could permit only one RW mount. Same iscsi volume mounted by multiple Pods
+			// conflict unless all other pods mount as read only.
+			if iqn == eiqn && lun1 == elun1 && target == etarget && !(volume.ISCSI.ReadOnly && existingVolume.ISCSI.ReadOnly) {
+				return true
+			}
+		}
+
 		if volume.RBD != nil && existingVolume.RBD != nil {
 			mon, pool, image := volume.RBD.CephMonitors, volume.RBD.RBDPool, volume.RBD.RBDImage
 			emon, epool, eimage := existingVolume.RBD.CephMonitors, existingVolume.RBD.RBDPool, existingVolume.RBD.RBDImage
@@ -150,6 +171,7 @@ func isVolumeConflict(volume v1.Volume, pod *v1.Pod) bool {
 // - GCE PD allows multiple mounts as long as they're all read-only
 // - AWS EBS forbids any two pods mounting the same volume ID
 // - Ceph RBD forbids if any two pods share at least same monitor, and match pool and image.
+// - ISCSI forbids if any two pods share at least same IQN, LUN and Target
 // TODO: migrate this into some per-volume specific code?
 func NoDiskConflict(pod *v1.Pod, meta interface{}, nodeInfo *schedulercache.NodeInfo) (bool, []algorithm.PredicateFailureReason, error) {
 	for _, v := range pod.Spec.Volumes {
diff --git a/plugin/pkg/scheduler/algorithm/predicates/predicates_test.go b/plugin/pkg/scheduler/algorithm/predicates/predicates_test.go
index 9932d6cd43..05c1502a57 100644
--- a/plugin/pkg/scheduler/algorithm/predicates/predicates_test.go
+++ b/plugin/pkg/scheduler/algorithm/predicates/predicates_test.go
@@ -738,6 +738,65 @@ func TestRBDDiskConflicts(t *testing.T) {
 	}
 }
 
+func TestISCSIDiskConflicts(t *testing.T) {
+	volState := v1.PodSpec{
+		Volumes: []v1.Volume{
+			{
+				VolumeSource: v1.VolumeSource{
+					ISCSI: &v1.ISCSIVolumeSource{
+						TargetPortal: "127.0.0.1:3260",
+						IQN:          "iqn.2014-12.server:storage.target01",
+						FSType:       "ext4",
+						Lun:          0,
+					},
+				},
+			},
+		},
+	}
+	volState2 := v1.PodSpec{
+		Volumes: []v1.Volume{
+			{
+				VolumeSource: v1.VolumeSource{
+					ISCSI: &v1.ISCSIVolumeSource{
+						TargetPortal: "127.0.0.2:3260",
+						IQN:          "iqn.2014-12.server:storage.target01",
+						FSType:       "ext4",
+						Lun:          1,
+					},
+				},
+			},
+		},
+	}
+	tests := []struct {
+		pod      *v1.Pod
+		nodeInfo *schedulercache.NodeInfo
+		isOk     bool
+		test     string
+	}{
+		{&v1.Pod{}, schedulercache.NewNodeInfo(), true, "nothing"},
+		{&v1.Pod{}, schedulercache.NewNodeInfo(&v1.Pod{Spec: volState}), true, "one state"},
+		{&v1.Pod{Spec: volState}, schedulercache.NewNodeInfo(&v1.Pod{Spec: volState}), false, "same state"},
+		{&v1.Pod{Spec: volState2}, schedulercache.NewNodeInfo(&v1.Pod{Spec: volState}), true, "different state"},
+	}
+	expectedFailureReasons := []algorithm.PredicateFailureReason{ErrDiskConflict}
+
+	for _, test := range tests {
+		ok, reasons, err := NoDiskConflict(test.pod, PredicateMetadata(test.pod, nil), test.nodeInfo)
+		if err != nil {
+			t.Errorf("%s: unexpected error: %v", test.test, err)
+		}
+		if !ok && !reflect.DeepEqual(reasons, expectedFailureReasons) {
+			t.Errorf("%s: unexpected failure reasons: %v, want: %v", test.test, reasons, expectedFailureReasons)
+		}
+		if test.isOk && !ok {
+			t.Errorf("%s: expected ok, got none.  %v %s %s", test.test, test.pod, test.nodeInfo, test.test)
+		}
+		if !test.isOk && ok {
+			t.Errorf("%s: expected no ok, got one.  %v %s %s", test.test, test.pod, test.nodeInfo, test.test)
+		}
+	}
+}
+
 func TestPodFitsSelector(t *testing.T) {
 	tests := []struct {
 		pod    *v1.Pod