This paper reports on a research study that focused on predicting the loss of construction workers' productivity due to thermal environment variations. The paper utilises a statistical polynomial regression analysis to establish a relationship between productivity and the predicted mean vote (PMV) thermal comfort index. In doing so, it builds upon a substantial amount of data reported in the literature regarding construction productivity and thermal environment. A set of equations reflecting the nature of the construction task being performed as well as the thermal environment are proposed to predict the degree of change in workers' productivity, according to a change in the thermal environment. The paper also reports on an experimental investigation undertaken to validate the developed equations. Validation results indicate that the developed equations can predict productivity with a reasonable level of accuracy. Furthermore, they show that the workers' productivity decreases as the PMV index moves away from the optimum range for all the observed tasks.