Replacement rates compare retirement income to pre-retirement earnings. They are used both by financing advisors in advising households on how much to save for retirement, and by policymakers who wish to assess the adequacy of retirement saving in the population. But in recent years, there has been renewed discussion of how replacement rates should be calculated. I discuss three topics in this current debate. First, should policymakers be presented with replacement rates based upon stylized workers or using administrative or microsimulation data? Second, should the denominator of the replacement rate calculation—pre-retirement earnings—be adjusted for the growth of prices or the growth of economy-wide wages? And third, should replacement rates incorporate a family-size adjustment to account for how having children affects parents’ need to save for retirement? I illustrate the effects of these methodological choices using a microsimulation model of Social Security benefits and employer-sponsored pensions. The results suggest that much of the disagreement over whether Americans face a ‘retirement challenge’ or a ‘retirement crisis’ stems not from differing estimates of how much income American retirees will have, but of how much income they will need to maintain their pre-retirement standards of living.