In the renewable energy production solid biomass has became one of the most important source for power and heat generation, in particular woody materials in the form of wood chips, pellet and briquette. Technical standards on solid biofuels require information about origin and source of the biomass, differentiating for example between coniferous and broadleaf. In this work different wood samples were classified employing a method based on thermogravimetric analysis followed by Principal Component Analysis and Soft Independent Modeling of Class Analogy as supervised pattern recognition method. The best results were obtained considering the temperature range between 200 and 300 degrees C, corresponding to hemicellulose degradation. The method results very efficient (100% recognition) at identifying between hardwood and softwood. Nevertheless it shows a good potential to classify single species. This method can be used to assess the quality of solid biofuels with respect to the requirements defined by the specific technical standards. (C) 2015 Elsevier Ltd. All rights reserved.