Based on a prescriptive theory of inductive reasoning, a training program to foster inductive reasoning has been developed. Children from 12 first-grade classes, mean age about 7 years, N = 279, participated in a training experiment. The children of 6 classes were trained to apply a strategy to reason inductively while the children of the remaining classes continued their regular classroom activities. It was expected that trained children would outperform the untrained children with respect to Raven's Coloured Progressive Matrices but not with respect to a vocabulary test, thus indicating convergent and discriminant or domain-specific training effects. Results confirmed this expectation. Moreover, it was expected that training would improve performance on the inductive subtests of Cattell's Culture Fair Test 1, but not influence subtests that did not involve inductive reasoning. Considerable transfer to both kinds of subtests was found on the immediate transfer task. However, with a delayed posttest 6 months later, the expected differential training effect could be observed. Finally, a LISREL model analysis confirmed the hypothesis that training children to reason inductively improved fluid but not crystallized intelligence.