OBJECTIVE To predict the cost of a delivery following assisted reproductive technologies (ART). DESIGN Cost analysis based on retrospective chart analysis. SETTING University-based ART program. PATIENT(S) Women aged >or=26 and <or=42 years with FSH levels <or=12 IU/L on day 3 undergoing a first cycle of fresh, nondonor ART. INTERVENTION(S) Logit regression using a fractional polynomial model of age and basal FSH was used to estimate the probability of a live birth. Cost analysis was applied to the resulting prediction. MAIN OUTCOME MEASURE(S) The predicted probability of a live birth following ART based on a woman's age and FSH and the associated cost of a live birth. RESULT(S) Analysis of 1,238 first ART cycles produced a prediction model for live birth rates following ART incorporating both age and FSH. A cost analysis based upon combination of age and FSH revealed the cost of a live birth exceeded $100,000 when the probability of a live birth fell below 15% and the cost rose exponentially at lower probabilities of live birth. CONCLUSIONS(S) Based upon a woman's age and FSH and expected cost for a live birth using ART may be calculated. At live birth rates <5%, the cost of ART is high and greatly exceeds the cost of donor cycles. This information is vital for patient counseling.