The robust laboratory evidence of preference reversal for lotteries has been interpreted as a threat to the general vailidity of standard theories of decision-making under uncertainty. This evidence is obtained from laboratory, that is, not real-world, lotteries with subjects who have not sought to make decisions among such lotteries. Here, the prevalence of preference reversal is studied in a field experiment with used cars, that is, a case of real-world non-trivial, non-lottery - but still payoff-uncertain - choice objects, and with subjects who registered as potential buyers of such cars. No sign of preference reversal was observed.