Purpose To determine the agreement and validity of keratometric measurements in children with the Nidek ARK-510A auto-refractokeratometer compared to rotating Scheimpflug imaging with Pentacam and biograph with Lenstar LS 900. Methods This study was conducted on 5620 schoolchildren aged 6–12 years in Shahroud, Iran. Minimum and maximum keratometry values and corneal astigmatism magnitude were compared by calculation of Paired difference, interclass correlation coefficient, and 95% limits of agreement (LoA) between devices. Results After applying the exclusion criteria, 4215 right eyes were enrolled into the study. Mean minimum keratometry with Nidek ARK-510A, Pentacam, and Lenstar was 43.13 ± 1.51, 43.14 ± 1.48, and 42.87 ± 1.46 diopters (D), respectively, and mean maximum keratometry was 43.97 ± 1.59, 44.00 ± 1.56, and 43.75 ± 1.54 D, respectively. Nidek ARK-510A overestimated minimum and maximum keratometry by 0.25 ± 0.37 and 0.22 ± 0.41, respectively, compared to Penatcam. The LoA between Nidek ARK-510A and Pentacam for minimum and maximum keratometry measurements were −0.98 to 0.47 D and −1.02 to 0.57 D, respectively. The LoA between Nidek ARK-510A and Lenstar for minimum and maximum keratometry measurements were −0.70 to 0.72 D and −0.79 to 0.85 D, respectively. The agreement between devices was best in emmetropes, worst in hyperopes. For astigmatic vector components, the agreements between devices were poor but best agreement was between Nidek ARK-510A and Pentacam. Conclusions Keratometry measurement with Nidek ARK-510A was not significantly different from Pentacam and Lenstar, and this device can be used in screening programs in emmetropes.