Humans might possess either a single (amodal) internal clock, or multiple clocks for different sensory modalities. Sensitivity could be improved by the provision of multiple signals. Such improvements can be predicted quantitatively, assuming estimates are combined by summation, a process described as optimal when summation is weighted in accordance with the variance associated with each of the initially independent estimates. We assessed this possibility for visual and tactile information regarding temporal intervals. In Experiment 1, 12 musicians and 12 nonmusicians judged durations of 300 and 600 ms, compared to test values spanning these standards. Bimodal precision increased relative to unimodal conditions, but not by the extent predicted by optimally weighted summation. In Experiment 2, six musicians and six other participants each judged six standards, ranging from 100 ms to 600 ms, with conflicting cues providing a measure of the weight assigned to each sensory modality. A weighted integration model best fitted these data, with musicians more likely to select near-optimal weights than non-musicians. Overall, data were consistent with the existence of separate visual and tactile clock components at either the counter/integrator or memory stages. Independent estimates are passed to a decisional process, but not always combined in a statistically optimal fashion.