PURPOSE: Our initial goal was to compare the efficacy and safety of a glaucoma drainage device and trabeculectomy for children with primary congenital glaucoma after angular surgery failure. However, we discontinued the study due to the rate of complications and wrote this report to describe the results obtained with the two techniques in this particular group. METHODS: This was a parallel, non-masked, controlled trial that included patients aged 0-13 years who had undergone previous trabeculotomy or goniotomy and presented inadequately controlled glaucoma with an intraocular pressure ≥21 mmHg on maximum tolerated medical therapy. We randomized the patients to undergo either placement of a 250-mm2 Baerveldt glaucoma implant or mitomycin-augmented trabeculectomy. The main outcome measure was intraocular pressure control. We calculated complete success (without hypotensive ocular medication) and qualified success (with medication) rates. We defined failure as uncontrolled intraocular pressure, presence of serious complications, abnormal increase in ocular dimensions, or confirmed visual acuity decrease. RESULTS: We studied 13 eyes of 13 children (five in the glaucoma drainage device group; eight in the trabeculectomy group). Both surgical procedures produced a significant intraocular pressure reduction 12 months after intervention from the baseline (tube group, 22.8 ± 5.9 mmHg to 12.20 ± 4.14 mmHg, p=0.0113; trabeculectomy group, 23.7 ± 7.3 mmHg to 15.6 ± 5.9 mmHg, p=0.0297). None of the patients in the tube group and 37.5% of those in the trabeculectomy group achieved complete success in intraocular pressure control after 12 months of follow-up (p=0.928, Chi-square test). Two patients (40%) had serious complications at the time of tube aperture (implant extrusion, retinal detachment). CONCLUSIONS: Both the tube and trabeculectomy groups presented similar intraocular pressure controls, but complete success was more frequent in the trabeculectomy group. Non-valved glaucoma drainage devices caused potentially blinding complications during tube opening. Because of the small sample size, we could not draw conclusions as to the safety data of the studied technique.