diff --git a/pkg/controller/podautoscaler/replica_calculator.go b/pkg/controller/podautoscaler/replica_calculator.go
index b065fce3743..c97c5f51140 100644
--- a/pkg/controller/podautoscaler/replica_calculator.go
+++ b/pkg/controller/podautoscaler/replica_calculator.go
@@ -105,9 +105,10 @@ func (c *ReplicaCalculator) GetResourceReplicas(ctx context.Context, currentRepl
 
 	if len(missingPods) > 0 {
 		if usageRatio < 1.0 {
-			// on a scale-down, treat missing pods as using 100% of the resource request
+			// on a scale-down, treat missing pods as using 100% (all) of the resource request
+			maxTargetUtilization := int64(max(100, int32(targetUtilization)))
 			for podName := range missingPods {
-				metrics[podName] = metricsclient.PodMetric{Value: requests[podName]}
+				metrics[podName] = metricsclient.PodMetric{Value: requests[podName] * maxTargetUtilization / 100}
 			}
 		} else if usageRatio > 1.0 {
 			// on a scale-up, treat missing pods as using 0% of the resource request
@@ -208,9 +209,10 @@ func (c *ReplicaCalculator) calcPlainMetricReplicas(metrics metricsclient.PodMet
 
 	if len(missingPods) > 0 {
 		if usageRatio < 1.0 {
-			// on a scale-down, treat missing pods as using 100% of the resource request
+			// on a scale-down, treat missing pods as using 100% (all) of the resource request
+			maxTargetUtilization := int64(max(100, int32(targetUtilization)))
 			for podName := range missingPods {
-				metrics[podName] = metricsclient.PodMetric{Value: targetUtilization}
+				metrics[podName] = metricsclient.PodMetric{Value: maxTargetUtilization}
 			}
 		} else {
 			// on a scale-up, treat missing pods as using 0% of the resource request