ALOHA was originally designed for communication between remote terminals and a central computer in Hawaii.For some time, however, multi-channel ALOHA, wherein ran domized channel selection replaces randomized retransmission delay upon collision, has been used in settings such as online transaction processing via geostationary satellites. There, the long propagation delay renders channel sensing impractical, and the short messages render reservation schemes useless. For these applications, striving to maximize attainable throughput while meeting a deadline with near certainty captures both the service provider's fixed costs and per-transaction revenue, the user's delay consciousness, and ALOHA's probabilistic nature.This paper reviews several techniques that we have developed for this purpose. Their common theme is devoting a large amount of channel resources to a message before allowing it to fail to meet the deadline, while keeping a low mean per message resource expenditure. This is achieved by assigning higher priority to messages in their later retransmission attempts.The performance improvement over naive schemes is dramatic, at times approaching the attainable unconstrained throughput.Finally, the schemes are practical.