Racial/ethnic bias in the prediction of students' educational potential was questioned in the Larry P. Vs. Riles case. The construct and predictive validity of the Woodcock-Johnson IV Tests of Cognitive Abilities (WJ IV; Schrank et al. 2014b) have not been examined for racial/ethnic bias. This study extended Keith's (1999) examination of bias across racial/ethnic groups for general and specific cognitive abilities predicting reading achievement. It employed a higher-order model of cognitive abilities based on Cattell-Horn-Carroll (CHC) theory and a three-factor reading achievement model adapted from Niileksela et al. (2016). Using samples of 9- to 13-year-old Black, Hispanic, and White children from the WJ IV norming sample, models representing CHC cognitive abilities and reading demonstrated invariance across groups. Despite the fact that the models where CHC broad abilities predicted reading achievement were a poor fit, results indicated similar predictive relations for Black children as for other racial/ethnic groups. The CHC cognitive models employed are not acceptable to predict reading for Black children--and all children--within this age range.