A study investigated the use of a mathematical model to predict individuals' total active Hebrew vocabulary from samples of their written and spoken language. The model is based on a generalized inverse Gaussian distribution. The subjects were Israeli junior high school students from both high and low socioeconomic groups. Hebrew language samples of at least 2,000 words were obtained from each child. In 60 of 70 cases, the fit between empirical data and the theoretical model was acceptable. Several hypotheses relating to the model's construct and concurrent validity and its reliability and objectivity were tested. In addition, a hypothesis about the significantly richer vocabulary of higher socioeconomic status children was confirmed. The model can serve as a viable basis for further extensive inquiry into the vocabulary of different samples. (Author/MSE)