PIP: The Luker model specifies that failure to contracept might be deliberate. The determinants of contraceptive behavior are the assignment of advantages and disadvantages to contraceptive use and pregnancy. A subjective probability is given to the likelihood of getting pregnant and to reversing any pregnancy that might occur. Successful risk taking leads to a lower probability of pregnancy and thus more risk taking. Stressful life events or interpersonal situations may change the probabilities. The advantage of this model is its proximity to the actual coital event. The goal is to identify factors related to risk taking. Examination of the model involved a sample of 425 sexually active women aged 13-19 years from a multiservice center for youth in New York City in 1981 who visited the agency at least 1 month prior to the interview. The sample was 36% black, 36% white, 24% Hispanic, and 4% Oriental or other. 77% were enrolled in school and 55% were from intact homes. 38% had professionally employed parents and 14% received welfare. Questions were directed to the likelihood of pregnancy and contraception and the advantages and disadvantages, the likelihood of abortion if pregnancy occurred, and background variables (social and psychological characteristics; situational factors at last intercourse; sexual, contraceptive and pregnancy histories; degree of support from partners, peers, parents; ego development; and knowledge of pregnancy risk and contraception). Reliability of each scale was assessed. The results showed that 26% of women at last coitus had taken a pregnancy risk. These risk takers considered birth control as more disadvantageous and pregnancy more advantageous. The probability of pregnancy at last coitus was considered by risk takers to be lower than non-risk takers. Risk and non-risk takers were at the same risk of pregnancy regardless of subjective perception. 50% thought about pregnancy or contraception at last coitus. 48% reported thinking about the good or bad things about getting pregnant, and 38% actually thought about being pregnant. 9 of 43 variables were related to risk taking. The multivariate models revealed that significant variables were knowledge of welfare history, previous risk taking, ego development, and 5 variables from the Luker model. Adding the scale for stressful life events did not affect the results. The Luker model and developmental reasons for pregnancy risk are supported. There was moderate support for Luker's feedback mechanism, but good longitudinal data are necessary for accurate cost accounting. The assumption that motivation to attend a clinic will affect all future coital acts is deceptive.^ieng