Increasing the “truth per dollar” of money spent on science is one legitimate long-run goal of the economics of science. But before this goal can be achieved, we need to increase our knowledge of the successes and failures of past and current reward structures of science. This essay reviews what economists have learned about the behavior of scientists and the reward structure of science. One important use of such knowledge will be to help policy-makers create a reward structure that is more efficient in the future.