The evidence for believing that mixtures of aspirin, phenacetin, and caffeine provide advantages over the individual components of these mixtures is reviewed, and doubt expressed as to the rationale for the use of these mixtures in ordinary medical practice. The syndrome of `analgesic nephropathy' is also reviewed, and on the basis of experiments in healthy volunteers it is suggested that individual ingredients of analgesic mixtures be scrutinized more carefully in an attempt to track down the agents responsible for toxic effects.