Summary: The random lottery incentive system is widely used in experimental economics to motivate subjects. This paper investigates its validity. It reports three experiments which compare responses given to decision tasks which are embedded in random lottery designs with responses in ‘single choice’ designs in which each subject faces just one task for real. The experiments were designed to detect cross-task contamination effects in the random lottery treatment. No significant differences between treatments, and no significant contamination effects, were found. Over the three experiments, observed differences between the treatments are adequately explained as sampling variation.