Published data on NiTi wire tensile tests display a surprising variety of results even though the same material has been studied. Hysteresis shapes can be observed that range from box- to cigar-like. In some cases, the variation may be the result of different post-fabrication treatment, such as annealing or cold working procedures. However, oftentimes local data are generated from average stress/strain concepts on the basis of global force and end displacement measurements. It is well known among experimentalists that this has a smoothening effect on data, but there is an additional, less well-known mechanism at work as well. This effect is due to thermomechanical coupling and the thermal boundary condition at the ends of the wires, and it manifests itself in a strong data dependence on the length of the employed specimen. This paper illustrates the effects of a thermal boundary layer in a 1D wire by means of an experimental study combined with a simulation based on the fully coupled momentum and energy balance equations. The system is modeled using COMSOL FEA software to simulate the distribution of strain, temperature, resistivity, and phase fractions. The local behavior is then integrated over the length of the wire to predict the expected behavior of the bulk wire as observed at its endpoints. Then, simulations are compared with results from a tensile test of a 100 mum diameter Dynalloy Flexinol wire between two large, steel clamps. Each step of the tensile test experiment is carefully controlled and then simulated via the boundary and initial conditions of the model. The simulated and experimental results show how the thermal boundary layer affects different length SMA wires and how the inhomogeneity prevents transition to austenite at the wire endpoints. Accordingly, shorter wires tend to be softer (more martensitic) than longer wires and exhibit a large reduction in recoverable strain because a larger percentage of their total length is impacted by the thermal boundary.