The Brailsford-Bullough model of recombination of point defects on the surface of coherent inclusions under irradiation is critically analysed. It is shown that the steady state solutions of the model exist only for relatively small coherent particles (of a few nanometres in size) and, depending on the configuration of the particle-matrix interface, at relatively low temperatures. For high temperatures, the model should be reformulated, taking into account thermal desorption of trapped vacancies, which makes it possible to explain the stabilization of trapped point defects in the steady state and to calculate the rates of removal of point defects by coherent inclusions.