Abstract Biomass burning is a significant source of particulate matter (PM) in ambient air. To date, the evaluation of the distinct contributions of both, wood burning, used for residential heating, and green waste burning, on PM concentration levels is difficult and rarely achieved. Such discrimination is of major concern for air quality policy-makers in order to implement effective actions to reduce air pollution. The main objective of the research project SODEMASS (bioMASS burning SOurces DEconvolution) is to identify specific organic molecular markers and/or chemical patterns of both biomass burning sources that can be further used in PM source apportionment studies. Several experiments have been performed in “real” conditions using different wood combustion appliances such as a residential wood stove and a fireplace. Several parameters have been investigated including the output conditions (nominal vs reduced) and the wood log moisture content (mix of species including beech, oak, and hornbeam). In addition, green waste burning experiments have been carried out in a large combustion chamber facility (1000 m3) using two kinds of burning material such as grass mowing with tree leaves and hedge trimming with branches. Several physicochemical parameters (smoke temperature, O2, NOx, CO, CO2, non-volatile PM concentrations) were monitored continuously by using automatic sensors or analyzers and about 50 PM samples (on quartz fiber filters) have been collected after dilution of the channeled smoke (dilution factor about 1000). Filter samples have been characterized using both, targeted (levoglucosan and its isomers, methoxyphenols, alkanes, polyols, catechols and derivatives as well as EC/OC) and non-targeted (high resolution mass spectrometry (Q-TOF) coupled to GC or LC) approaches. The data treatment and statistical analysis will allow highlighting specific organic molecular species (exact mass, retention time, molecular formula, MS/MS fragmentation…) or chemical pattern candidates of each biomass burning source.