Both spontaneous reports and single outcome studies may distort the overall safety evaluation of drugs. We identified epidemiologic studies, published from January 1970 to December 1995, that investigated the association of serious adverse effects with aspirin, diclofenac, acetaminophen, and dipyrone to determine and compare the excess mortality associated with short-term drug use. The estimated excess mortality due to community-acquired agranulocytosis, aplastic anemia, anaphylaxis, and serious upper gastrointestinal complications was 185 per 100 million for aspirin, 592 per 100 million for diclofenac, 20 per 100 million for acetaminophen, and 25 per 100 million for dipyrone. The estimates were largely influenced by the excess mortality associated with upper gastrointestinal complications. A relative risk estimate of 300 or more for the association of dipyrone with agranulocytosis would have been necessary for the excess mortality of dipyrone to be comparable to that of aspirin or diclofenac. Based on published epidemiologic evidence used to determine the excess mortality associated with short-term use of these four non-narcotic analgesics, the current regulatory ranking of the drugs appears inappropriate.