Recent evidence suggests reaching movements are more accurate when endpoint contact occurs, suggesting that finger tip contact contributes to a final estimation of arm position. In the present study we tested two hypotheses: (1) that fingertip contact influences illusions of arm movement produced by muscle vibration, and (2) that this influence depends on the a priori context of the stability of the contact surface. Subjects sat with their elbows on a table and eyes closed. They demonstrated the perceived orientation of their left (cue) arm by mirroring the location with their right (report) arm. We manipulated deep proprioceptive cues by vibrating the left biceps brachia, causing illusions of elbow extension, and tested whether these illusions were altered when the fingertip remained in contact with a stable external surface. The context at this point represents a prior assumption that the external contact surface is stable. Midway through the experiment the context was changed by challenging the prior assumption that the contact surface was stable by demonstrating that it could move. Unbeknownst to the subject, the external contact surface remained stable during data collection throughout the experiment. As expected, without tactile cues, biceps vibration caused illusory elbow extension. Conditions with fingertip contact and biceps vibration in the stable context demonstrated that contact largely eliminated the overestimation of cue arm elbow angle. However, in the context of a possibly unstable (movable) contact surface, the reports of elbow extension returned. Thus, a priori notions about the stability context of an external contact surface influence how this tactile cue is integrated with proprioceptive sensory modalities to generate an estimate of arm location in space. These findings support the notion that tactile cues are used to calibrate proprioception against external spatial frameworks.