The notion of a spatiotemporal window of integration is a commonly accepted concept in multisensory research: crossmodal information falling within this window is integrated, whereas information falling outside of this window is not. It has been recognized, however, that integrating crossmodal information always involves a decision about whether or not two (or more) sensory cues originate from the same event, i.e., have a common cause [e.g., Koerding et al. PLoSONE, Sept. 2007]. Several research groups have shown by now that multisensory integration more or less closely follows rules based on optimal Bayesian estimation procedures. Here we extend this approach by determining the optimal width of a time window of integration: An infinitely large time window would lead to mandatory integration, a zero-width time window would rule out integration entirely. Computation of an optimal time window must be based on, amongst others, the a-priori probability of a common cause and the likelihood of observed temporal disparities between the unimodal signals. We demonstrate this approach within the framework of the time-window-of-integration (TWIN) model developed by the authors [Colonius & Diederich, JCogNeurosci 2004; Diederich & Colonius, BrainRes 2008].