Forty patients with cancer pain receiving intermittent narcotics were admitted to a prospective study designed to assess the cognitive effects of narcotics. Twenty patients had undergone no change in narcotic dose or type greater than or equal to 7 days (stable dose, SD, group), and 20 patients had undergone an increase of greater than or equal to 30% in dose less than or equal to 3 days before (increased dose, ID, group). Age, primary tumor, type, dose and route of narcotic were not different between the SD and ID group. Cognitive tests (finger tapping, FT, 10 and 30 sec, arithmetic, 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 97 +/- 9%, 100 +/- 14%, 100 +/- 13%, 100 +/- 15%, 98 +/- 19%, in the SD group, vs. 77 +/- 14% (P less than 0.001), 83 +/- 13% (P less than 0.001), 124 +/- 21% (P less than 0.001), 60 +/- 21% (P less than 0.001) and 68 +/- 21% (P less than 0.001) in the ID group, respectively. Our results suggest that patients who undergo a significant increase in the dose of intermittent narcotics experience significant cognitive impairment, that disappears after 1 week of the increase. More research is needed to better characterize the cognitive toxicity of intermittent narcotics, and to determine the cognitive effects of long acting narcotics, continuous infusions, or of the addition of amphetamines.