This study compared the efficacy, safety and optimal dose of two formulations of cyclosporin, Sandimmun and Neoral, in patients with severe, chronic plaque-type psoriasis. Patients were randomized on a 1:1 basis to 24 weeks of treatment with Neoral (n = 152) or Sandimmun (n = 157). The starting dose of each formulation was 2.5 mg/kg per day. Dose increases to maintain efficacy were allowed after 4 weeks. In patients who achieved remission, the dose was down-titrated at 4-week intervals from week 16. The maximum permitted dose for each formulation was 5.0 mg/kg per day. Neoral produced a more rapid response than Sandimmun: remission rates were higher for Neoral during the first 8 weeks of treatment. The number of dose reductions for safety was similar in both treatment groups, but there were more dose increases to maintain efficacy in the Sandimmun group (198) than the Neoral group (146). The number of patients with dose reductions after week 16 was higher for Neoral (n = 83) than for Sandimmun (n = 73). The frequency and nature of adverse events were similar for both treatment groups. The mean dose required to control the disease was approximately 10% lower with Neoral and fewer dose changes were needed. The increased bioavailability and reduced pharmacokinetic variability of cyclosporin provided by the Neoral formulation may facilitate short-course, intermittent therapy.