This trial was a randomized, double-masked, crossover study during which patients with keratoconjunctivitis sicca underwent 6 weeks of treatment with either cyclosporine 1% ophthalmic ointment or placebo followed by 6 weeks of the alternative treatment. Washout periods using only unpreserved artificial tears preceded both treatment cycles. Twenty-five patients completed the first treatment period, but only eight met entry criteria for period II. Cyclosporine ointment was associated with initial mild to moderate redness, itching, and burning that returned to baseline levels within 1-2 weeks. Rose Bengal results and results of four subjective (patient diary) efficacy parameters favored cyclosporine: foreign body sensation, overall symptoms, hours of symptom control per day, and overall effectiveness. No systemic adverse events or laboratory abnormalities occurred. We conclude that (a) the crossover design is inappropriate for studying this disease; (b) mild to moderate itching, redness, and burning occur initially with cyclosporine administration, although tolerance quickly develops; (c) cyclosporine appears to benefit the ocular surface in keratoconjunctivitis sicca; and (d) further trials in this syndrome are warranted.