Abstract: Ten years ago, analgesics were studied using crossover designs. In recent years, analgesics have been studied only in parallel designs primarily because biostatisticians do not like crossover studies. The advantages of crossover studies are numerous: (1) patients serve as their own control; (2) there is less variability of responses among patients; and (3) a smaller number of patients is needed to provide statistically significant data. As long as crossover of treatment medications does not occur within 4 to 6 hours, the problem of carryover effect of the previous medication is insignificant or negligible. Two studies will be presented. One is a crossover study of Percodan with and without naloxone to placebo. The other is a parallel study comparing the effects of propoxyphene with naloxone to those of propoxyphene alone. The results of these studies reaffirm the value of the crossover method of evaluating analgesics.