Much research shows that when it comes to preferences over sequences of money, such as a monthly paycheck, people do not always maximize present value. Rather, they often choose the lower-valued of a pair of sequences, especially when it has attractive properties such as an increasing trend. To unearth the reasons for sequence preferences we conducted a verbal-protocol analysis of choices between money sequences, including lifetime and one-year earnings and one-year lottery winnings, as well as lifetime health sequences. Participants thought aloud while choosing between visual representations of sequences. Their verbalizations contained reference to a wide range of previously hypothesized, as well as new, reasons for choice. These reasons were also correlated in sensible ways with the choices made. There was some evidence of solid economic reasoning (which we called maximization), although this was largely restricted to choices for one-year earnings. More commonly, respondents did not distinguish between earnings and consumption, and thought about money as if they would automatically spend it at the rate it was received. This meant the most frequently given reason was what we called appropriateness-how well the money received at a given point matched the desired consumption at that point. Other common explanations include the expectedness of a sequence (we argue this is mediated by appropriateness), the way that parts of the sequence constitute reference points for other parts, and the ease with which incoming money can be managed (convenience). Copyright # 2002 John Wiley & Sons, Ltd. key words intertemporal choice; sequences; rationality; economic psychology; verbal protocolsGiven a choice between two ways of distributing a fixed amount of money over time, either as a falling sequence going from large payments to small, or a rising sequence going from small to large, a rational decision maker will choose the falling sequence. This is because at any non-zero rate of interest the falling sequence dominates the rising one, since any unspent money can earn interest. Despite this, there is evidence that people often prefer rising to falling sequences Chapman, 1996;Frank, 1992;Gigliotti & Sopher, 1997;Loewenstein & Sicherman, 1991;Matsumoto, Peecher, & Rich, 2000 1999). One real-world consequence of this is that workers demand rising wages (after inflation) even when their productivity is not rising, leading them to be underpaid early and overpaid late in their careers (Frank & Hutchens, 1993;Neumark, 1995). Loewenstein and Sicherman (1991) reported the first experimental demonstration of the preference for rising over falling sequences of money. Their participants ranked money sequences, and also chose between a single rising and falling one. For both tasks, the great majority chose rising over falling sequences. Several replications of this study, using slightly different materials, have found much the same pattern (Gigliotti & Sopher, 1997;Matsumoto et al., 2000;Susianto, 1999).Although the modal prefer...