diff --git a/cluster-autoscaler/core/static_autoscaler.go b/cluster-autoscaler/core/static_autoscaler.go
index d9c9fcf9b72b..772f2030e218 100644
--- a/cluster-autoscaler/core/static_autoscaler.go
+++ b/cluster-autoscaler/core/static_autoscaler.go
@@ -621,23 +621,27 @@ func (a *StaticAutoscaler) RunOnce(currentTime time.Time) caerrors.AutoscalerErr
 			a.processorCallbacks.disableScaleDownForLoop, scaleDownInCooldown)
 		metrics.UpdateScaleDownInCooldown(scaleDownInCooldown)
 
+		// We want to delete unneeded Node Groups only if here is no current delete
+		// in progress.
+		_, drained := scaleDownActuationStatus.DeletionsInProgress()
+		var removedNodeGroups []cloudprovider.NodeGroup
+		if len(drained) == 0 {
+			var err error
+			removedNodeGroups, err = a.processors.NodeGroupManager.RemoveUnneededNodeGroups(autoscalingContext)
+			if err != nil {
+				klog.Errorf("Error while removing unneeded node groups: %v", err)
+			}
+			scaleDownStatus.RemovedNodeGroups = removedNodeGroups
+		}
+
 		if scaleDownInCooldown {
 			scaleDownStatus.Result = scaledownstatus.ScaleDownInCooldown
+			if len(removedNodeGroups) > 0 {
+				a.processors.ScaleDownStatusProcessor.Process(autoscalingContext, scaleDownStatus)
+			}
 		} else {
 			klog.V(4).Infof("Starting scale down")
 
-			// We want to delete unneeded Node Groups only if there was no recent scale up,
-			// and there is no current delete in progress and there was no recent errors.
-			_, drained := scaleDownActuationStatus.DeletionsInProgress()
-			var removedNodeGroups []cloudprovider.NodeGroup
-			if len(drained) == 0 {
-				var err error
-				removedNodeGroups, err = a.processors.NodeGroupManager.RemoveUnneededNodeGroups(autoscalingContext)
-				if err != nil {
-					klog.Errorf("Error while removing unneeded node groups: %v", err)
-				}
-			}
-
 			scaleDownStart := time.Now()
 			metrics.UpdateLastTime(metrics.ScaleDown, scaleDownStart)
 			empty, needDrain := a.scaleDownPlanner.NodesToDelete(currentTime)