The effects of the Follow Through Program as determined by meta-analysis techniques were reviewed and summarized. The general conclusion that more direct instructional models had greater positive effects on measures of basic skills in mathematics and reading than less direct, developmental models was supported. However, the validity of the Follow Through Evaluation as a general test of the worth of instructional models was questioned by contrasting the Follow Through results with a similar meta-analysis of elementary school science process programs. The science process programs had substantial effects on the measures which matched their objectives--that is, science process measures--and minimal or negative effects on more distant objectives. It was concluded that testing a broad spectrum of outcomes, including attitude and creativity, is essential in evaluating a curricular instructional model. (Author/GDC)