We explore the small-scale spatial and temporal transferability of model parameters between two points in the ablation zone of tropical Glaciar Shallap, Cordillera Blanca, Peru (9°S, −77° W; ∼4800 m a.s.l.) in order to provide a robust assessment of the performance of a process-based glacier mass-balance model. Relative surface height change is calculated at hourly time-steps, and cumulative values are compared to surface height measurements made at irregular intervals (14–64 days) over the course of two continuous hydrological years (August 2006–August 2008). Best-performing parameter combinations were determined for each point from the outcome of 1000 model simulations for which parameters were varied randomly within a defined range. With these parameter combinations measurements for a specific location and time-span are well reproduced. Transferring the parameter combination as optimized for one location to the other location in the ablation zone increases the errors of modeled cumulative mass balance by 5–1326 mm ice eq.a−1. Transferring the parameter combinations as optimized for one year to the other year increases the modeled errors in cumulative mass balance by 18–3179 mm ice eq.a−1. Model errors generally increase during periods with frequent snowfall and snow cover. This could reflect either the inherent difficulty of modeling complex snow processes, or the inability of the model to correctly capture the pattern of albedo evolution at this site. The magnitude of errors associated with parameter transfer in space and time highlights the need for improving model performance for robust climatological and/or hydrological analyses on tropical glaciers.