Abstract: Pigeons were exposed to an analog of a `dual-task' procedure used to test attentional models of timing in humans. After separate training on an auditory duration discrimination and on a variable ratio (VR) schedule, VR episodes lasting for 5 s were superimposed on the stimuli to be timed, either early (E) or late (L) during the trial. Trials with VR yielded underestimation of the target durations (increased % of `short' choices), relative to trials without VR, and this effect was stronger under the L than under the E condition. Data were similar to those collected with humans and support attentional models of timing according to which the simultaneous non-timing task uses processing resources which are diverted from the timing mechanisms.