Objective: To determine if neuropsychological variables could be used as predictors of outcome in a substance abuse sample. Method: 61 participants (80% male, 76.9% African American) in their treatment program were selected by the Fulton County Accountability Court in Atlanta, GA to undergo neuropsychological assessment. All met criteria for Substance Abuse dependence. Participants had to be abstinent from all substances of abuse while in the program. The battery given to all participants included the Montreal Cognitive Assessment, Kaufmann-Brief Intelligence Test II, Wide Range Achievement Test-Revision 4, Rey-Osterrieth Complex Figure Test, Copy Phase, Finger Tapping Test, Hand Dynamometer, Trail Making Test (TMT) A & B, Beck Depression Inventory II and Beck Anxiety Inventory. The outcome variables studied were whether the participants graduated or were terminated from the program, and length of time in the program. Results: Among the portion of the sample who were terminated from the program, education level, R2 = .131, F(1, 34) = 5.136, p < .05; Trails A z-score R2 = .608, F(1, 8) = 12.409, p < .01; Trails A time, R2 = .693, F(1, 8) = 18.079, p < .05; Trails B z-score R2 = .561, F(1, 8) = 10.211, p < .05; and Trails B time, R2 = .649, F(1, 7) = 712.946, p < .01, predicted length in the program. Education was a significant moderator of this relationship. Conclusion(s): Performance on the TMT appears to predict time spent in the program and this effect is moderated by education level. A discussion of the cognitively mediated neuropsychological processes underlying successful performance on the TMT will be provided.