Twenty consecutive patients with cancer pain receiving a continuous subcutaneous infusion of narcotics were admitted to a double-blind, crossover trial designed to assess the effects of methylphenidate on neuropsychological functions. After a baseline assessment, patients were randomized to receive methylphenidate orally at 08.00 h for 2 consecutive days or placebo. During day 3, a crossover took place and patients received the alternate treatment for 2 more days. At the end of the trial, both patients and investigators were asked to choose blindly the most effective treatment. Cognitive tests (finger tapping, FT 10 and 30 sec; arithmetics, A; reverse memory of digits, RM; and visual memory, VM) were performed in all patients before and 45 min after their morning dose of narcotics for 2 consecutive days. Mean percentual change in FT 10 sec, FT 30 sec, A, RM, and VM after the narcotic dose were 89 +/- 12%, 98 +/- 15%, 108 +/- 17%, 101 +/- 17%, and 97 +/- 21% in the placebo group versus 119 +/- 16 (P less than 0.001), 130 +/- 19 (P less than 0.001), 78 +/- 21 (P less than 0.001), 125 +/- 28 (P less than 0.01), and 128 +/- 26 (P less than 0.001) in the methylphenidate group, respectively. After completion of the trial, methylphenidate was chosen blindly by the patient and investigator in 13 and 14 cases, placebo in 3 and 2 cases, a "no choice" in 3 and 3 cases, respectively (P less than 0.01). Our results suggest that methylphenidate is capable of improving cognitive function in patients receiving high doses of opiates subcutaneously. More research is necessary in order to determine the duration of cognitive improvement after each dose of methylphenidate as well as the best type and dose of amphetamine.