We explore the evolution of productivity and markups in U.S. food and beverage manufacturing from 1959 through 2018, paying particular attention to the latest 20 years. Using the Census-based NBER-CES Manufacturing Productivity Dataset, we estimate a translog production function which allows for Hicks-neutral and labor-augmenting technical change and measure the unobserved productivity growth. The results show that, in contrast to general manufacturing, productivity growth in food manufacturing has been driven more by Hicks-neutral productivity than by labor-augmenting productivity, and that the growth of productivity has been rather slow. Moreover, markups in food manufacturing have been more modest than found in previous studies of market power in the food industries and comparable to those of general manufacturing. In contrast to previous studies, we find no evidence of markups rising in either food manufacturing or general manufacturing in the last 20 years. We attribute three reasons by which these results contrast with the generalized beliefs about raising markups, that are simultaneously possible explanations for the puzzles that we detect in our data. These are: accounting problems that may explain the rise that we see before the 2000s: possible changes in the value of input elasticities associated with biased technological change that should be carefully considered; and aggregation of firm-level markups that should be done avoiding spurious correlation.