The relationship between costs and health benefits of branded pharmaceuticals remains controversial. This paper examines the incremental costs incurred for incremental health benefits gained from the largest available sample of cost-effectiveness studies of branded drugs in the USA, the 1994–2015 Tufts Registry of Cost-Effectiveness Analyses. Earlier studies used small, specialized samples of drugs. We use linear regression analysis to estimate the association in those studies between additional quality-adjusted life years (QALYs) and incremental pharmaceutical costs. The preferred sample uses 476 studies involving branded pharmaceuticals with both higher costs and increased effectiveness compared to the previous standard of care. Regressions of costs on QALYs imply that an additional QALY is associated, on average, with a $28,561 increase in cost (95 % CI, $18,853–$38,270). This regression explains 20 % of the variation in sample costs. In this analytical sample, a share of the variation in the cost of pharmaceuticals is, therefore, not random but rather associated with variation in QALYs; prices are to some extent “value-based.” Our results are robust to varying sample inclusion criteria and to the funding source. In subgroup analyses, the highest cost per QALY was $44,367 (95 % CI, $35,373–$53,361). Costs of pharmaceuticals in this data set are, on average, lower than common estimates of the monetary value of a QALY to American consumers. As in other studies, we find that sellers of patent-protected beneficial new technology appear to capture only a fraction of the benefits provided.