This study explains why gain/loss discount rate differences reported in previous studies cannot be attributed to outcome sign alone, but rather, must be associated with particular outcome-sign/question-frame combinations. To do so, it extends Loewenstein's (1988) framing model of intertemporal choice to negative outcomes and uses the resulting predictions to interpret and integrate the results of three previous studies comparing subjective discount rates (Thaler 1981, Loewenstein 1988, Benzion et al. 1989). The new framework reveals previously unidentified linkages among outcome signs, question frames, and discount rates. To investigate whether losses and gains are, in fact, discounted differently, an experiment is conducted that includes a neutral-frame intertemporal choice scenario (no proposed change in outcome timing) for each outcome sign. The results show that subjective discount rates vary in a predictable way according to the outcome sign and question frame combination examined.