A curriculum was devised by working backward from Stanford-Binet items to specification of a universe of content for which the Stanford-Binet could serve as a content-valid achievement test. It was reasoned that this curriculum should home. This curriculum was tested on 20 4-year-old disadvantaged children in Champaign-Urbana, Illinois. The program was conducted for eight months, two hours daily, with a teacher-pupil ratio of one-to-five. The Stanford-Binet was administered four times during the experiment, curriculum content and procedures being modified in the light of results. The Wechsler Preschool and Primary Scale of Intelligence (WPPSI) was administered at the beginning and end as a control measure for non-specific effects on IQ. Total IQ gain was 13 points on the Stanford-Binet--no better than that achieved previously with the highly structured programs which made no effort to teach Binet-related content. Gain from the WPPSI turned out to be of the same magnitude, thus indicating that the gains were in no wise test-specific. These results were taken as indicating that there was not a close relationship between curriculum content and intelligence test performance, leaving open the possibility that what accounted for the non-trivial part of the IQ gain might have been the accererated acquisition of certain basic thinking skills. (Author/MK)