Teenage pregnancy rates in the U.S. are among the highest in the world for industrialized countries. The generally accepted reason is not that American teenagers are more sexually active but that they contracept less than do teenagers in other industrialized countries. This dissertation reports on a study that was undertaken for two purposes. One purpose was to develop and test two models of contraceptive choice among American teenagers: a "likelihood-of-use" model to predict the likelihood of sexually active teenagers' using contraception, and a "medical-or-nonmedical" model to predict whether teenagers who use contraception are likely to use medical or nonmedical methods. The second purpose was to explore the level of support for the two models among black and white teenagers separately. The theoretical underpinning of the models is value-expectancy theory. The models' exogenous variables are based on the prevailing strategy for preventing teenage pregnancy among American teenagers, a strategy initially advocated by Joy G. Dryfoos. The strategy involves the use of access-to-contraception programs, educational programs, and life options programs. The data used in the study were on 449 subjects drawn from the 1979 National Survey of Young Women, a probability-sample survey of women in the U.S. aged 15-19. The subjects were those survey respondents who were black or white, sexually active, never married, and never pregnant. The statistical technique used in the study was logistic regression. Test results supported three of four hypotheses constituting the medical-or-nonmedical model and two of seven hypotheses constituting the likelihood-of-use model. The results for each model offered support for using two of the three programs constituting the prevailing pregnancy-prevention strategy: access-to-contraception programs and educational programs. Exploration of the level of support for each of the two models among black and white teenagers indicated that support for each model differed between the two groups of teenagers.