Partial comparative messages comprise a mixture of comparative and non-comparative claims. This study demonstrates how the extent to which non-comparative (NC), partial comparative (PC) and complete comparative (CC) messages influence consumer preference depends on the comparisons (i.e. NC, PC or CC) made by the messages promoting the two competing brands. A 3 (brand A used NC, PC or CC message) Â 3 (brand B used NC, PC or CC message) design was used. Two studies (n 5 284, 283, respectively) show that participants preferred (1) the first brand they saw when both brands were promoted by NC messages, (2) neither brand when both brands were promoted by messages containing comparative claims, (3) the brand promoted by a PC message when the other brand was promoted by an NC message, and (4) the brand promoted by an NC message when the other brand was promoted by a CC message.Comparative and non-comparative messages have received considerable attention (Belch, 1981;Chattopadhyay, 1998;Grewal et al., 1997;Muehling, Stoltman and Grossbart, 1990;Rose et al., 1993). Messages can be fully comparative, fully non-comparative or somewhere between these two extremes (Donthu, 1992). Numerous studies of comparative ads have failed to consider the share of message content devoted to comparison (Del Barrio-Garcia and Luque-Martı´nez, 2003;Hwang, 2002;Manning et al., 2001). The studies of Barone and Miniard (1999) and Barone, Palan and Miniard (2004) were exceptions because they utilized partial comparative (PC) messages. PC messages were defined as comprising a mixture of comparative and noncomparative claims. However, those studies did not examine the effects of PC messages in competition, demonstrating a knowledge gap in PC message effectiveness. Since competing messages are pervasive, this gap must be addressed.Numerous studies have investigated how messages promoting a single brand influence consumer retrieval of messages promoting other brands (Burke and Srull, 1988;Jewell and Unnava, 2003;Keller, 1991; Allen, 1993, 1994;Kent and Kellaris, 2001;Leong, Ang and Heng, 1999;Unnava and Sirdeshmukh, 1994). However, exactly how the comparative/ non-comparative nature of late-appearing messages promoting another brand influences consumer preferences for the brand promoted by the early-appearing message has seldom been addressed, showing a gap between academic study and practice. In practice, late-appearing messages are likely to affect consumer information processing regarding the early-appearing message, indicating the importance of research to fill this gap.To fill the two aforementioned gaps, this study investigates consumer preferences for two brands promoted via non-comparative (NC), PC or complete comparative (CC) messages. These messages contain straight non-comparative claims, a mixture of comparative and noncomparative claims, and straight comparative claims, respectively. This study has two unique features. First, the study examines the effective-