Precise timing is crucial to decision-making and behavioral control, yet subjective time can be easily distorted by various temporal contexts. Application of a Bayesian framework to various forms of contextual calibration reveals that, contrary to popular belief, contextual biases in timing help to optimize overall performance under noisy conditions. Here, we review recent progress in understanding these forms of temporal calibration, and integrate a Bayesian framework with information-processing models of timing. We show that the essential components of a Bayesian framework are closely related to the clock, memory, and decision stages used by these models, and that such an integrated framework offers a new perspective on distortions in timing and time perception that are otherwise difficult to explain.