The present paper explores empirically the relationship between teacher grading and student achievement. The hypothesis is that the teachers can manipulate student effort, and hence student achievement, by choosing the proper grading practices. The grading model is analogous to a labor supply model, where the teachers can set the marginal returns to achievement or determine the grade level that is independent of real achievement. The empirical analysis shows that grading differences in the lower secondary school in Norway are much like differences in non-labor income and, further, that students who are exposed to hard grading perform significantly better than other students.