PURPOSE: Episode-based cost sharing (EBCS), defined as prospectively set patients' cost-sharing obligations for an entire episode of care, is proposed to reduce cost uncertainty for patients. The aim of this study was to assess patient preferences for EBCS for three hypothetical care episodes: breast cancer screening (BCS), lung cancer screening (LCS), and childbirth. METHODS: Adult, English-speaking Amazon Mechanical Turk workers participated in three surveys, with eligibility varying on the basis of the screening eligibility criteria for BCS, LCS, and the ability to give birth. Using a choice-based experiment, participants stated their preferred cost-sharing models for BCS, LCS, and childbirth. Selections were from nine models constructed on the basis of two attributes each with three levels: (1) initial cost (none or low, intermediate, and high) and (2) subsequently revealed additional cost if other or different medical services were needed (none, intermediate, and uncertain). Data were analyzed using hierarchical Bayesian conjoint analysis to compare preferences for EBCS (high or intermediate initial cost, no subsequent cost for additional services) versus the status quo (no or low initial cost with uncertain cost of additional services). RESULTS: In total, 103, 103, and 102 individuals participated in the BCS, LCS, and childbirth surveys, respectively. Across all three surveys, EBCS models were preferred over the status quo by 18.1 percentage points (95% confidence interval [CI], 17.8-18.4 percentage points) for BCS, 12.0 percentage points (95% CI, 11.7-12.3 percentage points) for LCS, and 44.5 percentage points (95% CI, 44.1-44.9 percentage points) for childbirth. Cost-sharing models with uncertain additional costs were the least preferred across all surveys. CONCLUSIONS: Health care consumers prefer cost-sharing models that prospectively guarantee out-of-pocket costs for entire care episodes.