High-resolution radio observations have revealed that non-thermal radio emission in WR stars arises where the stellar wind of the WR star collides with that of a binary companion. These colliding-wind binary (CWB) systems offer an important laboratory for investigating the underlying physics of particle acceleration. Hydrodynamic models of the binary stellar winds and the wind-collision region (WCR) that account for the evolution of the electron energy spectrum, largely due to inverse Compton cooling, are now available. Radiometry and imaging obtained with the VLA, MERLIN, EVN and VLBA provide essential constraints to these models. Models of the radio emission from WR146 and WR147 are shown, though these very wide systems do not have defined orbits and hence lack a number of important model parameters. Multi-epoch VLBI imaging of the archetype WR+O star binary WR140 through a part of its 7.9-year orbit has been used to define the orbit inclination, distance and the luminosity of the companion star to enable the best constraints for any radio emitting CWB system. Models of the spatial distribution of relativistic electrons and ions, and the magnetic energy density are used to model the radio emission, and also to predict the high energy emission at X-ray and gamma-ray energies. It is clear that high-energy facilities e.g. GLAST and VERITAS, will be important for constraining particle acceleration parameters such as the spectral index of the energy spectrum and the acceleration efficiency of both ions and electrons, and in turn, identify unique models for the radio spectra. This will be especially important in future attempts to model the spectra of WR140 throughout its complete orbit. A WCR origin for the synchrotron emission in O-stars, the progenitors of WR stars, is illustrated by observations of Cyg OB2 No. 9.