A model of attrition for specially admitted black students (low high school achievement and low assessed college aptitude) at the University of Pittsburgh was tested. The majority fell into the lower end of a socioeconomic range and often had inadequate precollege preparation. The attrition model was examined using a linear structural equation technique called LISREL (Joreskog, 1979), which estimates unknown coefficients in a set of equations and specifies hypotheses in terms of cause and effect variables and their indicators. The following variables were assessed: family background, type of high school attended, entry ability, on-campus academic behavior, and attrition. For the 80 black students entering the university in 1976-1977 and 1977-1978 through the special admission program, significant relationships were found among the following variables: entry ability, high school attended, on-campus academic behavior, and attrition. The construct most highly related to attrition was on-campus academic behavior. The important indicators of on-campus academic behavior were the adequate and timely completion of homework, regular class attendance, asking questions in class, careful and complete studying for tests, and putting forth maximum effort for class success. Entry ability was found to be significantly related to the type of high school attended and attrition. Scholastic Aptitude Test (SAT) scores for verbal and math abilities were found to be effective measures of entry ability. Information is provided on: the range of SAT scores of the sample, range of high school ranks, degree expected, family income, parent educational background, and type of high school attended by the study group. Schemata of the model are included. (SW)