ANN's are computer algorithms capable of robust classification, pattern recognition, and the solution of other optimization problems characterized by complex combinations of variables. The ANN models consist of layers of interconnected nodes patterned after biological neurons in the brain and have the ability to model nonlinear, highly correlated functions for which a direct algebraic solution may not be feasible. The current research used one of the most commonly used ANN's, called feed-forward or multilayer perceptron with backpropagation. Feed-forward perceptrons, one of the first ANN's to be described and used as a linear computational tool, consist of two layers of nodes and a single layer of interconnections. This study assessed adolescent marijuana use and the clinical features of dependence based on self-evaluation from recent National Household Surveys on Drug Abuse. The effect of training and testing the neural networks with randomly selected data was compared to data selected as a function of the survey year. The study's specific aim was to account for adolescent marijuana use and features of marijuana dependence based on experiences with alcohol and tobacco use. The similarities in the performance of multiple logistic regression (MLR) and ANN model suggests there may be no major complex or nonlinear relationships in cross-sectional epidemiological data selected for adolescent drug use and dependence in this specific application. The authors advise that ANN's be further studied in future longitudinal research, possibly with the modeling of recursive networks so as to allow feedback from drug dependence to levels of marijuana use. They also indicate that ANN models have the potential to model drug use and dependence based on input parameters with no obvious direct link to drug use. This would permit the identification of higher risk youth by using assessments indirectly related to adolescent drug use and dependence. 4 tables, 6 figures, and 37 references