We performed trapezoidal keratotomy, consisting of combined nonintersecting semiradial and transverse incisions, in 64 eyes of 45 consecutive patients with naturally occurring astigmatism. The central clear zone diameter and number and length of transverse incisions were determined by the refractive error. Mean preoperative refractive astigmatism was 3.18 +/- 1.16 diopters (D) (range, 2.25 to 7.00 D). At the 1-year follow-up examination, the mean surgically corrected astigmatism determined by vector analysis was 3.70 +/- 1.50 D (range, 0.75 to 8.5 D), and the mean residual refractive astigmatism was 0.85 +/- 0.72 D (range, 0 to 4.0 D), with 64% of eyes having 1.00 D or less. The smaller the clear zone diameter, the greater the astigmatic correction. Longer transverse incisions produced more steepening of the secondary meridian. The operative complications included microperforation (5%), misalignment of surgical meridian (6%), encroachment on clear zone (5%), and inadvertent crossed incisions (11%). Trapezoidal keratotomy reduced naturally occurring astigmatism, but with only fair predictability and with some irregular astigmatism due to irregular wound healing.