The number of drugs approved by the FDA for treating cancer has increased substantially during the last 40 years. Moreover, cancer drug innovation has been accelerating: more than 8 times as many new cancer drugs were approved during 2005-2015 as were approved during 1975-1985 (66 vs. 8). During the period 2010-2014, the average annual growth rate of cancer drug expenditure was 7.6%--more than 3.6 times the average annual growth rate of nominal U.S. GDP. This has contributed to a lively debate about the value and cost-effectiveness of new cancer drugs. In this study, we attempt to assess the average cost-effectiveness in the U.S. in 2014 of new cancer drugs approved by the FDA during 2000-2014. Cost-effectiveness is measured as the ratio of the impact of new cancer drugs on medical expenditure to their impact on potential years of life lost due to cancer. We use a difference-in-difference research design: we investigate whether there were larger declines in premature mortality from the cancers that had larger increases in the number of drugs ever approved, controlling for the change in cancer incidence and mean age at time of diagnosis. The vast majority of the data we rely on are publicly available.