Diffusion imaging coupled with fiber tractography (DFT) measures the course of fibers through living tissue.  These measurements have been demonstrated for fascicles or bundles of fibrous cells within central nervous tissue, cardiac muscle and skeletal muscle.  Despite the success of current DFT methods, many limitations still exist.  Our work focuses on a unique application of DFT in which we are confident that two anatomical regions are connected and we wish to use DFT to identify the most likely pathways between these regions. For instance, in brain tissue we may know that homologous visual areas on both hemispheres of cortex have a direct connection or with muscle we may know the origin and insertion  locations and simply want to explore the most likely connections between these regions.  We introduce a probabilistic DFT algorithm (MetroTrac) to achieve this goal. We introduce the algorithm in three parts: a scoring procedure to measure the likelihood of a pathway, a sampler to systematically explore the set of possible pathways, and an inferential step that specifies how the sampled pathways can be used to derive a conclusion.  Our approach is demonstrated with multiple human brain cases as well as lower extremity muscles for one human subject.