Abstract To address the limitations of self-report measures, we developed the Pro-Environmental Behavior Task (PEBT) as a computerized paradigm for the assessment of actual pro-environmental behavior under controlled laboratory conditions. On each PEBT trial, participants can either choose the faster car option, which causes a series of lights to be illuminated, or they can save the associated energy by choosing the bicycle option at the expense of spending more time in the laboratory. In two pre-registered studies (both N = 120), we showed that the proportion of environmentally friendly PEBT choices is a valid and reliable measure of pro-environmental behavior. PEBT choices were consistent across trials, correlated to conceptually relevant variables, and sensitive to conceptually relevant manipulations. These effects were replicable and independent of the labelling of PEBT options. Our findings highlight the psychometric quality and utility of the PEBT as a paradigm that can open new avenues for research on pro-environmental behavior.