Current smart thermostats aim to increase the efficiency of heating and cooling systems by adjusting the temperature whenever the conditioned zone becomes empty. However, targeting energy savings, these systems often fail to achieve a comfortable thermal environment for the inhabitants. We propose to increase thermal comfort by automatically monitoring the inhabitants' satisfaction with the thermal environment using commodity hardware. To this end we designed the Comfstat infrastructure and publish detailed temperature and heart-rate data of seven users of the system to the community. Using this data we show that thermal comfort can be inferred automatically from a combination of sensor data within 0.5 points on the ASHRAE scale.