The cognitive processes students use in doing the 1996 science assessment of the National Assessment of Educational Progress (NAEP) were studied using two booklets from the 1993 NAEP science field test. Blocks of items from these booklets, a hands-on task block and either a conceptual/problem solving block or a theme block, were administered to 16 eighth graders who varied in proficiency in science as measured by the Metropolitan Achievement Tests. Students were interviewed after the test, and they were informed that the purpose of the study was to understand their thought processes as they answered the test items. Participants were offered a gift certificate incentive to complete the test. The combination of think-aloud protocols and concurrent interviews following the assessment was an effective way to explore participants' thought processes. Evidence suggests that the assessment is tapping the constructs it was designed to assess, namely science knowledge structure, reasoning, and hypotheses formulation and testing. In addition to lack of opportunity to learn, other main reasons for item difficulty were lack of factual knowledge, lack of conceptual understanding, and lack of knowledge of principles. The effects of motivation and test anxiety on performance were also apparent. (Contains 2 tables, 11 figures, and 17 references.) (SLD)