In a multicentre crossover study of 97 patients with mild hypertension, the incidence and severity of adverse effects were observed during the first 14 days of treatment with amlodipine, nifedipine retard or placebo. Amlodipine (5 mg) once daily was equipotent to nifedipine retard (20 mg) twice daily. At these doses, the incidence of adverse effects was significantly greater during treatment with nifedipine retard (41%) than with amlodipine (27%, P < 0.05) or placebo (16%, P < 0.01). In particular, headache and flushing occurred significantly less frequently during the first 14 days of treatment with amlodipine than with nifedipine retard. The lower incidence and reduced severity of vasodilatory side-effects associated with amlodipine resulted in fewer withdrawals during initiation of therapy (2 on amlodipine compared with 7 on nifedipine retard).