OBJECTIVE: We developed an evidence-based decision aid to help women with a prior cesarean to prioritize their childbirth preferences related to a future birth. Because there was uncertainty about which scale format would assist the patients in being most consistent in prioritizing preferences in a multiattribute decision model, we compared a graphic-numeric scale with a text-anchored scale. STUDY DESIGN AND SETTING: Ninety-six postnatal women with a prior cesarean were randomized to use 1 of 2 preference scale formats in a computerized childbirth decision aid. We measured the level of inconsistency (intransitivity) when patients prioritized their childbirth preferences and clarity of values before and after using the decision aid. RESULTS: When the trade-offs involved risk, women were more consistent when using graphic-numeric than text-anchored formats (P=0.015). They prioritized safety to their baby as 4 times more important than any other decision factor including safety to self. Both groups reduced unclear childbirth values over time (P<0.001). Women who over-used the extreme ends of the scale when evaluating risk were more likely to be inconsistent (P<0.001). CONCLUSION: Patients were more consistent in making trade-offs involving risk using graphic-numeric formats than text-anchored formats to measure patient preferences.