This research examines the effects of question order on the output of a customer satisfaction model. Theory suggests that locating product attribute evaluations prior to overall evaluations of satisfaction and loyalty should increase the impact of performance drivers in the model, explain more variation in the overall evaluations, and make positive satisfaction and loyalty evaluations more extreme. Our results show that, although customers′ overall evaluations are more extreme and better explained when provided after attribute evaluations, the impact of satisfaction drivers is relatively unaffected. Consistent with expectations, question order does affect the explained variation in satisfaction and the levels of satisfaction and loyalty. Implications for satisfaction modelling are discussed.