Broadcasting is an effective way of delivering data to a large population.' In the broadcast environment under consideration, a server broadcasts data items to all clients simultaneously, according to a certain transmission schedule. Users with pending data requests need to listen to the broadcast channel until their requests are satisfied by the transmitted data. Past research on broadcast scheduling assumes that once a user starts to wait for some data item, the user waits until the desired data item is transmitted by the server. This is often not true in practice. For various reasons, users may lose patience after waiting "too long" and leave with their requests unserved. In this paper, we study the broadcast scheduling problem taking user impatience into account. Based on our analytical results, we propose a scheduling algorithm that can produce a broadcast schedule with high service ratio (i.e., percentage of requests served) as well as low mean waiting time for the requests. Performance evaluation results based on simulations are provided.