Abstract. We introduce a novel Monte Carlo method for the analysis of optical probe design that couples a forward and an adjoint simulation to produce spatial-angular maps of the detected light field within the tissue under investigation. Our technique utilizes a generalized reciprocity theory for radiative transport and is often more efficient than using either forward or adjoint simulations alone. For a given probe configuration, the technique produces rigorous, transport-based estimates of the joint probability that photons will both visit any specified target subvolume and be detected. This approach enables the entire tissue region to be subdivided into a collection of target subvolumes to provide a phase-space map of joint probabilities. Such maps are generated efficiently using only one forward and one adjoint simulation for a given probe configuration. These maps are used to identify those probe configurations that best interrogate targeted subvolumes. Inverse solutions in a layered tissue model serve to illustrate and reinforce our analysis.