OBJECTIVES: Cost-efficacy analysis (CEA) (e.g., cost per cure or short-term response) is increasingly being used as an alternative or a supplement to cost-utility analyses (CUA) employing the incremental cost per quality-adjusted life-year (QALY) gained. The objective of this study was to investigate whether conclusions drawn from such CEAs were consistent with those of the gold standard CUA. METHODS: We developed a model comparing standard of care (SOC) with a new drug. At the end of therapy, patients incurred short-term costs and either achieved a response or not. Non-responders incurred additional long-term costs and lost additional QALYs compared with responders. We evaluated two scenarios. In Scenario A: the short-term cost was $40,000 and efficacy was 65% with SOC. Drug cost and efficacy were $50,000 and 90%, respectively. For responders, the long-term cost offsets were $10,000 and additional QALYs were 0.5. Scenario B differed from Scenario A in only two aspects: the drug was more costly ($60,000), and the additional QALYs were higher (2.0 QALYs). We computed average cost-efficacy ratio (ACER), incremental cost-efficacy ratio (ICER), and incremental cost-utility ratio (ICUR). The assumed threshold for cost-effectiveness was $50,000/QALY. RESULTS: We found that a lower ACER than SOC was neither necessary (Scenario A: $66,667 vs $61,539/responder and ICUR=$35,000/QALY) nor sufficient (Scenario B: $55,556 vs $61,539/responder and ICUR=$60,000/QALY) for a more efficacious drug to be considered cost effective. Although the drug had a higher ICER in Scenario A ($80,000 per additional responder) than Scenario B ($40,000/responder), the drug was cost-effective in Scenario A, and not so in Scenario B. We derived a formula that related ICER to ICUR. CONCLUSIONS: A lower average or incremental cost-efficacy ratio from a CEA was neither a necessary nor a sufficient condition for a new drug to be considered cost effective compared with SOC based on a CUA.