OBJECTIVES: Compare the cost effectiveness of 4 chemotherapy treatments for platinum-sensitive recurrent epithelial ovarian carcinoma (EOC). METHODS: A Markov model was constructed using a hypothetical cohort of 500 women (median age 60) to compare 4 NCCN recommended treatment-regimens for platinum sensitive recurrent EOC:  carboplatin/paclitaxel (C/P); carboplatin/gemcitabine (C/G), C/G with bevacizumab (C/G+B); and carboplatin/pegylated liposomal doxorubicin (C/PLD).  These treatments were chosen as they are each supported by phase III trials.  An indirect treatment comparison methodology was used to obtain evidence of the difference in treatment effects of each regimen.  Progression free survival (PFS) and overall survival (OS) data were used for survival comparisons.  The time horizon was thirty years.  Cost calculations were based on data from Medicare and published literature, and were based on median cycle number from each trial.  Published values of health utilities were used for QALY calculations.  Cost effectiveness ratios (CER) were calculated for each regimen, and expressed as 3 incremental cost effectiveness ratios (ICER):  additional month PFS, month OS, and QALY.  Reported rates of grade 3/4 toxicities from each trial were added to the cost of each treatment.  Cost, survival, and toxicity rate were varied over a range for sensitivity analysis.  RESULTS: C/G was a cost-effective regimen.  The cost for treating 1 woman with 6 cycles of C/G ranged from $1,140 (no toxicity) to $7,030  (toxicities at the reported rate).  Treatment with C/G produced a dominant ICER of $236,318/month-PFS. For each PFS-month gained over the next most cost-effective option, over $200,000 was saved.  C/G was the dominant strategy for OS, (ICER=$72,213/month OS).  When adjusted for health utility, C/G was the dominant strategy (ICER of $20,443/QALY). CONCLUSIONS: C/G was a cost-effective regimen, resulting in a dominant ICER for PFS, OS, and QALY.  C/G resulted in a savings compared to the next most cost effective regimen.