The current presentation of financial statements provides scant guidance for constructing the relation between revenues and operating costs. This relation is central for the analysis of profitability and earnings variability, which are key factors in generating operating earnings predictions and in assessing risk. Moreover, commonly used profitability ratios imply that operating earnings are proportional to sales. This paper infuses structure into the analysis of profitability and earnings variability by disaggregating operating costs into three components: a revenue-driven component, a capital-driven component, and a component that does not vary with either revenue or capital. This framework clarifies the relation among revenues, operating costs, and operating earnings. Because cost structure is not observable, I develop a firm-specific time-series estimation procedure that incorporates economic cost theory and accounting conventions with statistical analysis. I then examine the usefulness of cost structure information in (1) predicting next-period operating earnings, (2) assessing both systematic and idiosyncratic risk, and (3) explaining stock prices, while controlling for other information. The results indicate that cost structure information has implications for earnings forecasting, risk assessment, and consequently for equity valuation.