This paper develops and applies a model in which doctors have two dimensions of skill: diagnostic skill and skill performing procedures. Higher procedural skill increases the use of intensive procedures across the board, while better diagnostic skill results in fewer intensive procedures for the low risk, but more for the high risk. Deriving empirical analogues to our theoretical measures for the case of C- section, we show that poor diagnosticians can be identified in the data and that improving diagnostic skill would reduce C-section rates by 15.5% in the bottom half of the risk distribution, and increase them by 5.5% in the top half. Such an change in the allocation of procedures would improve birth outcomes among all women.