The reported success rates in the treatment of congenital glaucoma with goniotomy, trabeculotomy, and trabeculectomy suggest that trabeculectomy should be performed if the other procedures fail. We propose that the decision to perform primary trabeculectomy in primary and secondary congenital glaucoma reduces the effect which the many variable findings in surgical anatomy may have on the outcome of other procedures. This is a retrospective study of the results of primary trabeculectomy in 21 consecutive eyes of 15 patients with congenital glaucoma. Eighteen of 13 patients' eyes were controlled after a single trabeculectomy and remained controlled after a mean follow-up of 3.9 years (range 1.5 to 6.7 years). The role of primary trabeculectomy in congenital glaucoma merits further consideration.