The present study contributes to the opportunities in producing gaseous or liquid substitution fuels by biomass pyrolysis or gasification. By both processes, the issued fuel characteristics are quite different so that the application fields have to be optimized. Using a predictive model, pyrolysis mass balance could be evaluated so that the fuel qualities could be predicted. In the present study, a predictive model based on proximate analysis of biomass input has been applied on experimental results from three agricultural residues (olive tree cuttings, rapeseed residues and soya residues) which have been pyrolyzed and gasified in a captive sample (fast pyrolysis) and a fixed bed reactor respectively. The experimental results were further compared with the results from the predictive model. The comparison showed deviations ranged between 0% and 28%. The highest deviation appeared in the net calorific value of the char. For the gasification results, on the other hand, the model does not fit the experimental results due to the differences in the assumptions taken. It is only used for the prediction of the heating value of the produced gas. (C) 2011 Elsevier B.V. All rights reserved.