This paper addresses both the construct validity and the criterion-related validity of the "Revisie Amsterdamse Kinder Intelligentie Test" (RAKIT), which is a cognitive ability test developed for primary school children. The present study compared immigrant primary school children (N = 559) and Dutch children (N = 604). The mean scores of Surinamese/Netherlands Antillean, Moroccan, and Turkish children differed from each other and were lower than those of the Dutch children. Comparison of the test dimensions showed that group differences with respect to the construct validity were small. We found some item bias, but the combined effects on the sum score were not large. The estimate of general intelligence (g) as computed with the RAKIT showed strong predictive validity for most school subjects and standardized achievement tests. Although some criteria revealed significant prediction bias, the effects were very small. Most of the analyses we performed on differences in test scores and differences in criterion scores supported Spearman's hypothesis that g is the predominant factor determining the size of the differences between two groups. The conclusion that the RAKIT can be used for the assessment of groups from various backgrounds seems warranted.