An air-conditioning control evaluation apparatus includes a storage unit and a computing unit. The storage unit stores building information, input information, control information, a set of building models, and a candidate selection criterion. The computing unit determines an item available as input data for a building model, identifies the distribution of observed data, selects a plurality of candidate building models from the set of building models based on the available item and candidate selection criterion, estimates each parameter based on a method corresponding to the distribution, determines one building model based on a predetermined statistic calculated for the plurality of building models and the residual between estimated and observed values calculated for each of the building models, and evaluates, by use of the determined building model, energy saving and comfort for a plurality of controls to be evaluated.