Delay discounting (the devaluation of rewards delayed in time) has been studied extensively using animal models with psychophysical adjustment procedures. Similar procedures were soon developed to assess delay discounting in humans. Although across species the same mathematical function relates discounted value to imposed delay, several methodological concerns have been implicated in human delay discounting procedures. A procedure recently developed to address these concerns is the Experiential Discounting Task (EDT). This task arranges experienced delays and rewards that humans make decisions regarding—experiencing the outcomes of their choices within session before making additional choices. The popularity of this procedure has been fueled by reports of its sensitivity to acute experimental manipulation, and that it has been predictive of treatment success. Similar sensitivity results have not been found when a traditional delay discounting task (DDT) has been used. Though the EDT appears useful for a variety of reasons, it has not been subjected to the same rigorous internal validity and reliability tests that traditional DDTs have. In two experiments we examined the test-retest reliability of the EDT (Experiment 1) and the way in which choice trials are regulated (Experiment 2). Results demonstrate that the EDT is reliable across time and choice is insensitive to trial regulation differences. We conclude with a critique of the EDT as a procedure for assessing delay discounting and hypothesize other processes it may be measuring.