CONTEXT: Improved continuation of oral contraceptive use may decrease the incidence of unintended pregnancy. Therefore, identifying predictors of continuation is important to informing targeted interventions.

METHODS: A survey conducted in 2003–2005 collected data from young women who sought oral contraceptives at three family planning clinics in New York, Dallas and Atlanta. Cluster analysis of data from 1,245 women was used to identify three distinct groups on the basis of baseline responses to items measuring constructs from the transtheoretical model of health behavior change: perceived advantages of pill use, perceived disadvantages and self-confidence in ability to sustain use. Logistic regression was used to assess the likelihood of continuation among the clusters.

RESULTS: Participants who were confident in their ability to use the pill and who considered both its advantages and its disadvantages important were more likely to continue use for six months than were those who gave low scores to the method’s advantages and their ability to use it (odds ratio, 1.4). The same appeared to be true for those who gave high scores to the pill’s advantages and their ability to use it, but a low score to the method’s disadvantages (1.3); however, this finding was only marginally significant. Demographic variables were not significant after adjustment for cluster membership.

CONCLUSIONS: Cluster membership based on transtheoretical model constructs helps predict oral contraceptive continuation. Women who undervalue the advantages of pill use and have low confidence in their ability to use the method represent a target for interventions and future research.