Proportional reduction is a common cartel practice in which cartel members reduce their output proportionately. We develop a method to quantify this reduction relative to a benchmark market equilibrium scenario and relate the reduction to the traditional conduct parameter. Our measure is continuous, allowing us to have an intuitive interpretation as the “degree of collusion” and nesting the earlier models in the existing literature. Furthermore, our methodology addresses Corts’ (1999) critique by estimating time-varying degree of collusion from a short panel of firm-level observations, exploiting firms’ ex post heterogeneity. We illustrate the method using the Joint Executive Committee railroad cartel data.