A matrix test was devised to assess cognitive skills associated with inferential reasoning. The format of the test requires the subject to choose, from among four alternatives, a figure which is related to other given figures on the basis of appearance, content, or spatial position in the matrix. The test consists of three perceptual matching items, 18 class membership items, 11 one- way classification items, and 12 two-way classification items. Only minimally dependent upon the communication of verbal instructions and requiring no verbal respons s from the subject, the test is suitable for use with young and disadvantaged children. To identify the factors that influence performance on the matrix test, comparative studies have been conducted with (1) lower and middle class urban populations from kindergarten and grades one, two, and three, (2) 4-year-old children from various Head Start programs and from a middle class nursery school program, and (3) 5-year-old children from Mount Olive, North Carolina and from Rome, Georgia. These studies have shown that the matrix test is a useful tool for obtaining data relevant to the early education of disadvantaged children. Refinement of the instrument and further analysis of the available data are projected. The project report contains detailed results of the various studies and includes five statistical tables. (JS)