The author studies the effects of capital reallocation (the flow of productive capital across firms and establishments mainly through changes in ownership) on aggregate labour productivity. Capital reallocation is an important activity in the United States: on average, its total value is 34 per cent of U.S. GDP. Firms with lower productivity are more likely to be reallocated to (i.e., bought by) more productive firms. Reallocated establishments experience an increase in productivity. The author develops a dynamic model of capital reallocation and compares its predictions with U.S. data. In the model, limited participation in acquisition markets by heterogeneous firms results in an increase in aggregate productivity. With reasonably chosen parameter values, policy experiments show that the increased reallocation of capital and labour contributed as much as a 17 per cent improvement in aggregate labour productivity in the mid-1980s. When a positive total-factor-productivity shock occurs, in steady state the increase in aggregate productivity arises entirely from this shock, and reallocation is unaffected.