To compare the outcomes of two different cross-linking (CXL) protocols, in pediatric keratoconus eyes. In this prospective randomized contralateral eye interventional study, 68 eyes of 34 patients, 9-16years old, underwent CXL and enrolled between October 2011 and October 2013. Group A represents conventional riboflavin-ultraviolet A (UVA)-induced CXL with 30min of exposure to UVA irradiation of 3mW/cm2. Group B represents accelerated cross-linking with 5min of continuous UVA irradiation of 18mW/cm2. In either group, total energy delivered was adjusted to 5.4J/cm2. Follow-up of all patients was accomplished throughout the postoperative 3years, and the data from the preoperative, 12, 24, and 36months visits were analyzed and compared in both groups. Uncorrected visual acuity, corrected distance visual acuity, steepest keratometry (Kmax), corneal astigmatism (simulated K), total wavefront aberrations, central corneal thickness (CCT), corneal densitometry, manifest refraction spherical equivalent, and endothelial cell density (ECD) were evaluated at baseline, 12, 24, and 36months post-CXL. At 1-year assessment, the mean value of UCVA, CDVA, and Kmax showed a statistically significant difference between both groups, without any documented change in the variables throughout the remaining follow-up (1-3-year) period. Twelve months postoperatively, mean LogMAR UCVA and CDVA were (0.11 ± 1.60) and (0.03 ± 1.60), respectively, in accelerated CXL group, compared to conventional CXL group values of (0.20 ± 1.00) and (0.06 ± 1.22), showing a statistically significant difference (P < 0.05). Mean Kmax in accelerated CXL group (45.47 ± 0.44) showed a statistically significant difference (P < 0.05) compared to conventional CXL group (46.41 ± 1.59) at 12months post-CXL. On the other hand, wavefront aberrations, simulated K, corneal densitometry, ECD, and CCT changes showed nonstatistically significant difference in conventional CXL group, compared to accelerated CXL group (P > 0.05) throughout the follow-up course. Both conventional and accelerated CXL improved UCVA and CDVA, attenuated disease progression, and reduced corneal steepness and wavefront aberrations at 1, 2, and 3years postoperatively. In no case did keratoconus progress over the 36-month follow-up.