Studies about thermal comfort in mixed-mode buildings have been performed in order to better understand this type of building and its influence on occupants' thermal perception. However, there is still no consensus amongst researchers regarding whether mixed-mode buildings should be evaluated separating each mode of operation (natural ventilation or air-conditioning) and whether adaptive thermal comfort theory applies to both modes of operation. Does the mode of operation of a mixed-mode building, ceteris paribus, influence occupant thermal comfort perception? Trying to answer such questions, field studies on thermal comfort were conducted in three mixed-mode office buildings in the city of Florianopolis (a temperate and humid climate), Southern Brazil. Buildings were equipped with mechanical cooling systems and operable windows, both controlled by occupants. Thermal comfort questionnaires were collected at the same time and location that environmental variables were measured by microclimate instruments. Almost 5500 questionnaires were answered by occupants of the three buildings in both modes of operation over the four seasons. Analysis of the results indicated that occupants' thermal perception was influenced by the mode of operation. Adaptive thermal comfort models were developed for natural ventilation and air-conditioning mode of mixed-mode buildings. This work found no evidence to support a single adaptive model for mixed-mode buildings. During natural ventilation mode, occupants adapted to indoor temperature fluctuations as predicted by the adaptive thermal comfort theory. On the other hand, during air-conditioning operation a weak adaptive relation (indoor comfort temperature vs. outdoor climate) was observed - a range of about 4 degrees C of indoor temperature fluctuation may be used for the operation of the air-conditioning system without compromising thermal comfort, which could help saving energy. This work is a first step towards building an adaptive model of thermal comfort for Brazilian subtropical climate. (C) 2017 Elsevier B.V. All rights reserved.