This study aimed to propose an optimization process for a free-form building shape in terms of the thermal load characteristic in the early design stage. Geometric modeling of a model free-form building was performed using a parametric design method with Rhinoceros. The model free-form building's surface was divided into finite elements by generating a mesh using Grasshopper, which is an add-in program of Rhinoceros. Geometric information was extracted from each finite element and used to estimate the heat gain and loss characteristics of the whole free-form building. A free-form building shape optimization process was proposed based on the genetic algorithm (GA). Its applicability was demonstrated by deriving the optimized shape of the model free-form building for various climate zones. Established models that returned the thermal characteristics of a free-form building were used as objective functions, which are critical in the GA optimization process. The results showed that the proposed process could rapidly predict and optimize the variation of the heat gain and loss characteristics that was caused by changing the building shape. (C) 2014 Elsevier B.V. All rights reserved.