Two recent studies proved the usefulness of predictors of academic performance and conceptual models that were developed primarily from research on traditional college students for the study of retention among nontraditional adult students in conventional postsecondary programs. Some modifications and extensions were suggested. The first study investigated the accuracy with which conventional demographic and achievement measures would differentiate between persisters and nonpersisters among adult students at Youngstown State University. A stepwise discriminant analysis identified four variables that explained 25 percent of the variance in the study population: financial aid, grade point average, matriculation status, and age. Using these four variables in the discriminant function to classify persisters and nonpersisters resulted in the correct classification of 81 percent of the study population--88 percent of the nonpersisters and 57 percent of the persisters. The second study explored the appropriateness of the conceptual model developed by Tinto from research on dropout among traditional college students for understanding dropout among a nontraditional student group (older, mostly minority women on welfare) enrolled in a postsecondary nondegree vocational training program. Findings indicated program completion was related not to personal background characteristics but to academic aptitude and the significance of students' integration to both academic and non-academic sectors. (YLB)