Walberg's (1981) nine-factor productivity model was tested with a national probability sample of 3,116 adolescents. A three-wave longitudinal design incorporated data from students, teachers, and parents to construct a latent-variable model of science achievement. A revised model provided a better fit than five alternative specifications, including a direct-influence model used in 23 previous studies. The revised model was also supported by cross-validation on a split-half subsample. Eight productivity factors contributed in diverse ways to science achievement. Prior achievement, peer environment, and the amount and quality of instruction had positive direct effects. Instructional time also proved instrumental in the achievement process by mediating the effects of other factors, including mass media, peer environment, and class environment. Home environment and motivation had the greatest indirect effects, particularly mediated through prior achievement. The implications for improving educational performance are discussed.