Recent years have witnessed a debate over whether to use the incremental cost-effectiveness ratio (ICER) or the incremental net health benefit (INHB) as the basis for comparative cost-effectiveness analyses. The ICER is simple and grounded in economic theory, but difficult issues arise in its estimation from study data. The INHB is more difficult to interpret but easier to estimate. The choice of methods is thus posed as a tradeoff between simplicity and statistical tidiness, but that is to misread the facts, as the statistical and economic issues are inextricably intertwined. Moreover, in a sense ICER and INHB are two sides of the same coin, linked by Fieller's method for the construction of confidence intervals for a ratio parameter. I will review the two approaches, discuss their properties, and argue for a synthetic view that acknowledges their strengths and pitfalls.