Taking into account the ambiguities and limitations of prevailing statistical models, such as losing data related to complicated and nonlinear interactions between psychological constructs and some of the assumptions like homogeneity of variances and normal distribution, the present research investigated the capability of Artificial Neural Networks Model for con ducting predictive studies. A sample of 456 male senior high school students responded to the California Personality Inventory (CPI; Gaff, 1975) and Adjustment Inventory for School Students (AISS; Sinha & Singh, 1993), and was categorized into five levels of adjustment (from maladjusted to completely adjusted). Factor analysis of various combinations of personality traits suggested that some of the networks could not predict adjustment due to non conformity between the number of variables and network architectures. However, a re- vision of the architectures and repetition of new networks significantly increased the proportion of correct predictions (the proportion of participants categorized into the indicated levels of adjustment based on AISS). The most appropriate network for predicting adjustment included a combination of the cognitive variables of flexibility, femininity, communality and tolerance.