diff --git a/Assets/Mirror/Transports/Latency/LatencySimulation.cs b/Assets/Mirror/Transports/Latency/LatencySimulation.cs
index 0277d7960..f4adf6af6 100644
--- a/Assets/Mirror/Transports/Latency/LatencySimulation.cs
+++ b/Assets/Mirror/Transports/Latency/LatencySimulation.cs
@@ -22,10 +22,34 @@ struct QueuedMessage
 
     [HelpURL("https://mirror-networking.gitbook.io/docs/transports/latency-simulaton-transport")]
     [DisallowMultipleComponent]
-    public class LatencySimulation : Transport
+    public class LatencySimulation : Transport, PortTransport
     {
         public Transport wrap;
 
+        // implement PortTransport in case the underlying Tranpsport is a PortTransport too.
+        // otherwise gameplay code like 'if Transport is PortTransport' would completely break with Latency Simulation.
+        public ushort Port
+        {
+            get
+            {
+                if (wrap is PortTransport port)
+                    return port.Port;
+                
+                Debug.LogWarning($"LatencySimulation: attempted to get Port but {wrap} is not a PortTransport.");
+                return 0;   
+            }
+            set
+            {
+                if (wrap is PortTransport port)
+                {
+                    port.Port = value;
+                    return;
+                }
+                
+                Debug.LogWarning($"LatencySimulation: attempted to set Port but {wrap} is not a PortTransport.");
+            }
+        }
+
         [Header("Common")]
         // latency always needs to be applied to both channels!
         // fixes a bug in prediction where predictedTime would have no latency, but [Command]s would have 100ms latency resulting in heavy, hard to debug jittering!