A double-blind, 3-phase, cross-over, placebo-controlled trial of the pain-relieving effect of amitriptyline and carbamazepine was carried out in 15 patients with central post-stroke pain (CPSP) but without signs of depression. Treatment was given, in randomized order, for periods of 4 weeks, separated by 1 week wash-out. The final doses were 75 and 800 mg/day, respectively, for amitriptyline and carbamazepine. The treatment effects were assessed by daily ratings of pain intensity on a 10-step verbal scale and at the end of each treatment period by a global rating of the analgesic effect on a 5-step verbal scale. For the assessment of depression the Comprehensive Psychopathological Rating Scale (CPRS) was used. Amitriptyline produced a statistically significant reduction of pain when compared to placebo. According to the global rating, 10 of the 15 patients were responders to this drug. The effect could already be noticed during the second treatment week and it appeared to be correlated to the plasma concentration, since the median total ami- and nortriptyline concentrations were 497 and 247 nmol/l, respectively, for responders and non-responders. The early onset, together with the fact that the patients were not depressed, nor did they obtain reduced scores on ratings of depressive symptoms and signs, provides strong support for the conclusion that the pain relief was not caused by an antidepressive effect. Five of the 14 patients treated with carbamazepine reported some pain relief, but the effect did not reach statistical significance when compared to placebo. No correlation was found between effect and plasma concentration. In general, the patients tolerated the planned final dose of amitriptyline well. No final dose reduction was necessary. Carbamazepine caused more side effects and the final dose had to be reduced in 4 patients. However, only 1 patient had to be taken off medication, on day 25, due to drug interaction.