Twenty-two patients (8 male, 14 female) with refractory epilepsy entered a balanced, double-blind, placebo-controlled crossover trial of nimodipine as adjunctive therapy. Treatment periods of 12 weeks (nimodipine 30 mg tds, 60 mg tds, 90 mg tds each for 4 weeks and matched placebo) were followed by wash-out intervals of 4 weeks. Five patients withdrew (2 side-effects, 1 intercurrent illness, 2 non-compliance). Median values (placebo vs. nimodipine) did not vary for total (17 vs. 18), partial (14 vs. 18) and generalised tonic-clonic seizures (2 vs. 5) or seizure days (13 vs. 13). Monthly analysis also failed to uncover any benefit for nimodipine. Side-effects were reported no more frequently with nimodipine than with placebo and pulse and blood pressure did not alter significantly. Antiepileptic drug levels were not affected by nimodipine treatment but circulating nimodipine concentrations were low. In this trial, nimodipine did not fulfil the promise of its success in animal models of epilepsy. Enzyme induction by concurrent antiepileptic therapy may provide an explanation.