Abstract. Temperature-index models have been widely used for glacier-mass projections spanning the 21st century. The ability of temperature-index models to capture non-linear responses of glacier surface mass balance (SMB) to high deviations in air temperature and solid precipitation was recently discussed in the context of mass-balance simulations employing advanced machine-learning techniques. Here, we performed numerical experiments with a classic temperature-index model and confirmed that such models are capable of detecting non-linear responses of glacier SMB to temperature and precipitation changes. Non-linearities derive from the change in the degree-day factor over the ablation season and from the lengthening of the ablation season.