The present thesis employed psychophysical choice procedures to examine the role of the cerebellum in interval timing in Wistar rats within the context of W. H. Meck and R. M. Church's (1983) mode control model of timing and counting. Three competing hypotheses were examined: That the cerebellum functions as 1) an independent milliseconds timer; 2) part of an extended timing system where it contributes to scalar variance; and 3) part of an extended timing system where it contributes to constant rather than scalar variance. Counting was also evaluated as a way to examine a specific source of constant variance in timing. A review of the mode control model and sources of variance in counting and timing, together with consideration of the generalised Weber function, concluded that a single set of processes could accommodate performance across the milliseconds and seconds range timing as well as counting. Lesions to the cerebellar hemispheres but not the cerebellar vermis produced some deficits in a millisecond discrimination task (200 to 800 ms) but discriminations in the seconds range (2 to 8 s) were unaffected by either type of lesion. In contrast, comparative lesions to nucleus accumbens produced deficits in both time ranges. Cerebellar hemisphere lesions but not vermal lesions also produced deficits in numerical discrimination. These findings suggest that damage to the cerebellar hemispheres influences a source of constant variability, because constant variability is a prominent source of error during millisecond timing but is masked by other sources of variability when tinting longer durations (> 2 s). The deficits in numerical discrimination suggest that switch processes, a specific source of constant variance described by the mode control model, are disrupted by damage to the cerebellar hemispheres. Prior to the lesion work, an extensive examination was also made of timing and numerical performance to establish that intact rats could discriminate the numerosity of trial unique signals which obviate concerns about non-numerical confounds. This work provided an unequivocal demonstration that rats can count sequential events, but they do so according to H. Davis and J. Memmott's (1983) "last resort" hypothesis.