This paper aims to estimate the cost allocation coefficients of agricultural activities based on incomplete information. A model to disaggregate the variable input costs based on the maximum entropy theory was developed. It was applied to the Alentejo region using data from the FADN network of 2008. Results show that entropy theory is a flexible and robust option for estimating cost allocation coefficients of agricultural activities in a context of incomplete information. Between model variants used, the model of generalized cross entropy had the best forecast capacity, once the estimated coefficients are closer to the real values.