The numerous reports on preference reversals in preference elicitations pose a great challenge to empirical economics. Many studies have found that different procedures may generate substantially different preferences. However, little is known about whether one procedure is more susceptible to preference reversals than another. Therefore, taking the preference reversals as a robust behavioral pattern, guidelines are called for to provide directions regarding a preferred preference elicitation task. This paper puts forward a new test of the internal consistency of choice and matching tasks, based on “internal preference reversals”. We replicate the preference reversal phenomenon and find a significant higher consistency within choice tasks than within matching tasks.