A design is described that uses need for supplemental (rescue) analgesic as a factor predicting effectiveness of a test analgesic. This methodology is especially suited for evaluating long-acting analgesics given repeatedly. Rescue use is measured over dosing intervals as test drug is titrated from a subanalgesic dose to that requiring no or minimal rescue. This design was used to evaluate oral long-acting morphine sulfate (MS Contin) given every 12 hours in a crossover study of cancer pain using oral immediate-release morphine sulfate given every 4 hours as reference. Less morphine was required for MS Contin given every 12 hours relative to immediate-release morphine sulfate given every 4 hours (186 +/- 22 mg vs. 239 +/- 35 mg; p = 0.04). Total daily morphine for both regimens correlated linearly (r = 0.96) with a slope of 1.27 +/- 0.11, significantly (p = 0.03) different from equivalence (slope of unity) in favor of MS Contin. This design features assay sensitivity (dose-response) and provides relative potency estimates for analgesics given at specific regimens.