This paper studies, both theoretically and empirically, the optimal executive compensation when firm performance is a noisy signal of executive’s hidden effort and the volatility of firm performance is stochastic. We build a tractable dynamic principal-agent model and show analytically that pay-for-performance sensitivity (PPS) decreases in both short-run and long-run components of volatility, but the mechanisms are different: 1) the short-run volatility directly affects the effort level implemented by the optimal contract, and 2) the long-run volatility affects PPS through its impact on firm value sensitivity to the short-run volatility. Using the short-run and long-run volatilities estimated from stock option data, we find evidence supporting the model’s implications: while both short-run and long-run volatilities are negatively correlated with PPS, firm operating performance, a direct outcome of executive effort in our model, ONLY decreases with the short-run volatility. Our paper highlights the importance, as well as the intricacies, of stochastic volatility in executive compensation design.