A two-year study that attempted to measure patterns of pupil attributions for success and failure in mathematics is reviewed. An attribution pattern variable (ATTPAT) was created to categorize individuals studied into three groups. The first group had an attribution pattern thought to lead to generalized low expectancies for success. Students classified as "high" were in the second level of ATTPAT, and pupils placed in the third level were those who could not be categorized within either of the first two groups. It was found that a student's perceptions about how others viewed him constituted important causal variables in both years of the study. Interest in and the rated importance of mathematics by students was heavily affected by performance in the second year of the study only. In general, the results and analysis indicated that notions about the functions of attribution patterns may need to be seriously revised. Factors identified as calling for further study include more explicit specification of hypotheses, models, and predictions about the causal role of attributions, with particular regard to the direction and magnitude of effects. Neither the grade level of the students nor the mathematics being studied is specified. (MP)