Two studies are reported where people are asked to accept or not a price reduction on a target product. In the high (low) relative saving version, the regular price of the target product is low (high). In both versions, the absolute value of the price reduction is the same as well as the total of regular prices of planned purchases. As first reported by Tversky and Kahneman (1981), findings show that the majority of people accept the price discount in the high-relative saving version whereas the minority do it in the low one. In Study 1, findings show that the previous preference reversal disappears when planned purchases are strongly related. Also, a previously unreported preference reversal is found. The majority of people accept the price discount when the products are weakly related whereas the minority accept when the products are strongly related. In Study 2, findings show that the classic preference reversal disappears as a function of the comparative price format. Also, another previously unreported preference reversal is found. When the offered price reduction relates to a low-priced product, people are more inclined to accept it with a control than a minimal comparative price format. Findings reported in Studies 1 and 2 are interpreted in terms of mental accounting shifts. Copyright © 2002 John Wiley & Sons, Ltd.