In the 1st reported experiment, we demonstrate that auditory memory is robust over extended retention intervals (RIs) when listeners compare the timbre of complex tones, even when active or verbal rehearsal is difficult or impossible. Thus, our tones have an abstract timbre that resists verbal labeling, they differ across trials so that no "standard" comparison stimulus is built up, and the spectral change to be discriminated is very slight and therefore does not shift stimuli across verbal categories. Nonetheless, performance in this nonverbal immediate memory task was better at short (1-, 2-, or 4-s) than long (8-, 16-, or 32-s) RIs, an outcome predicted by temporal distinctiveness theory whereby at long RIs, tones are closer in time to tones on previous trials. We reject this account in the 2nd experiment, where we demonstrate that the ratio of RI to intertrial interval makes absolutely no difference to performance. We suggest that steady forgetting is consistent with a psychoacoustically derived conception of an auditory memory (the "timbre memory model") that embodies time-based forgetting in the absence of feature-specific interference. (Contains 1 table, 7 figures and 2 footnotes.)