The paper discusses partitions in asynchronous message-passing systems. In such systems slow processes and slow links can lead to virtual partitions that are indistinguishable from real ones. This raises the following question: what is a "partition" in an asynchronous system? To overcome the impossibility of detecting crashed processes in an asynchronous system, our system model incorporates a failure suspector to detect (possibly erroneously) process failures. Based on failure suspicions we give a definition of partitions that acccounts for real partitions as well as virtual ones. We show that under certain assumptions about the process behavior, any incorrect failure suspicion inevitably partitions the system. We then show how to interpret the "absence of partition" assumption.