To assess the accuracy of various Pentacam indices in distinguishing keratoconus (KC) in pediatric patients with high astigmatism and to establish appropriate cutoff values. This prospective multicenter cross-sectional study included 312 eyes from 167 patients aged 6 to 18 years (mean age, 13.1 ± 3.2 years) evaluated with Oculus Pentacam HR. Patients were categorized into 4 groups: KC, forme fruste keratoconus, astigmatism greater than 2 diopters (Cyl2D), and control. A subgroup of Cyl2D comprised patients with astigmatism greater than 4 diopters (Cyl4D). Twenty-three Pentacam indices were analyzed, and receiver operating characteristic curves determined optimal cutoff points, sensitivity, and specificity. The best indices for distinguishing KC from Cyl2D were high-order aberration root mean square of the anterior corneal surface (area under the receiver operating characteristic curve 0.987), Belin/Ambrosio enhanced ectasia total derivation (0.971), index of vertical asymmetry (0.971), average pachymetric progression index (0.962), maximum Ambrosio relational thickness (0.960), posterior elevation (0.952), and anterior elevation (0.948). The accuracy of these indices was highest in the control group and lowest in the Cyl4D group. Area under the receiver operating characteristic curve was significantly lower for fruste keratoconus than KC. Optimal cutoff values were higher for astigmatic patients than for those with no refractive error. Most indices effectively distinguished between KC and normal pediatric patients. For individuals with high astigmatism, we suggest focusing on specific indices such as high-order aberration root mean square, Belin/Ambrosio enhanced ectasia total derivation value, maximum Ambrosio relational thickness, average pachymetric progression index, and index of vertical asymmetry. Optimal cutoff points for these patients were higher than those for nonastigmatic children and differed from adult populations.