Ion-exchange is a popular method to strengthen alkali-containing glasses. However, the experimentally observed residual compressive stress profiles often deviate from the theoretical profiles: a subsurface compression maximum is often observed, while monotonically decreasing stress with depth is expected. It is suggested here that this discrepancy can be accounted for by considering a surface stress relaxation process that takes place at a faster rate than bulk stress relaxation in the presence of water. This suggestion is based upon experimental data indicating the presence of water impurity in the molten salt used for ion-exchange strengthening of glasses. Recent work characterizing surface stress relaxation in oxide glass fibers as a diffusion controlled process aided by molecular water diffusion was applied to better explain the development of a subsurface compressive maximum within an accelerated timeframe. By combining the new surface stress relaxation diffusion mechanism with the well-established ion-exchange mechanism, a mathematical model predicting a subsurface maximum stress is developed with a good agreement to published data. •A surface relaxation mechanism is shown to cause subsurface compressive stress peak.•Hastened surface relaxation is due to trace water impurity in the molten salt.•Analytical models developed to predict ion-exchanged glass stress-depth profiles.