Abstract : Reservoir planning without the explicit accommodation of evaporation loss leads to errors in capacity estimates. However, whenever evaporation loss is considered, its quantification uses linear approximations of the intrinsically nonlinear height–area–storage (H–A–S) relationship to estimate the reservoir area, leading to bias in capacity estimates. In this work, biases resulting from using various H–A–S models are evaluated. These models include linear and nonlinear functions, either specifically developed for the case-study sites or available in the Global Reservoir and Dam (GRanD) database. All empirically derived approximations used data for two dams in India: the Bhakra on Sutlej River and the Pong on the Beas River, both tributaries of the Indus River. The results showed that linear H–A–S models underestimate the exposed surface area of the Pong reservoir by up to 11.19%; the bias at Bhakra was much less. The GRanD H–A–S model performed very poorly at both reservoirs, producing overprediction in exposed reservoir area of up to 100% and 415% at the Pong and Bhakra reservoirs, respectively. Analyses also showed that up to 29% increase in reservoir capacity is required to compensate for the effect of net evaporation loss at low demand levels. As demand increases, the required evaporation-correction capacity decreases in proportional terms and is indistinguishable for all H–A–S models. Finally, recommendations are made on using the results for evaporation adjustment at nongauged sites in the region.