Yes and no. You can for sure implement actors which support cancellation or killing them, the only question is how. In lots of actor frameworks you send the actor a close message that triggers it to shut down - and you can also send it a cancel message after some request. However if the server side actor is not implemented fully asynchronously and doesn't read it's messages, it might take a long time for the cancellation to happen -> exactly the same time which it takes for the operation to finish. E.g. if the remote actor is implemented in some sequence like: receiveRequest(); doSomethingWithRequestWhichIsProbablyBlockingRequestsToOtherActors(); sendResponse(); then there is no cancellation possibility. If the other actor is implemented fully asynchronously then it could pick up the cancellation and do something with it. But it's harder to implement, becaue then you have again state machines everywhere. In erlang killing actors without messages might work, because it's supported deeply within the runtime.