A model of the mass loss of Camembert-type cheese was established with data obtained from 2 experimental ripening trials carried out in 2 pilot ripening chambers. During these experiments, a cheese was continuously weighed and the relative humidity, temperature, oxygen, and carbon dioxide concentrations in the ripening chamber were recorded online. The aim was to establish a simple but accurate model that would predict cheese mass changes according to available online measurements. The main hypotheses were that 1) the cheese water activity was constant during ripening, 2) the respiratory activity of the microflora played a major role by inducing heat production, combined with important water evaporation, 3) the temperature gradient existing inside the cheese was negligible, and the limiting phenomenon was the convective transfer. The water activity and the specific heat of the cheeses were assessed by offline measurements. The others parameters in the model were obtained from the literature. This dynamic model was built with 2 state variables: the cheese mass and the surface temperature of the cheese. In this way, only the heat transfer coefficient had to be fitted, and it was strongly determined by the airflow characteristics close to the cheeses. Model efficiency was illustrated by comparing the estimated and measured mass and the mass loss rate for the 2 studied runs; the relative errors were less than 1.9 and 3.2% for the mass loss and the mass loss rate, respectively. The dynamic effects of special events, such as room defrosting or changes in chamber relative humidity, were well described by the model, especially in terms of kinetics (mass loss rates).