Abstract The objective of this experimental study consists in investigating the effects of torrefaction treatment on two different woody biomass chosen from softwood as spruce pine and annual grass lignocellulosic hardwood as reed . This thermal pre-treatment has been conducted in a small batch reactor directly fluxed by a controlled nitrogen flow. Tests have been carried out at different temperatures, ranging from 250 °C to 310 °C, and reaction time from few minutes to 1 h. Properties involved in this investigation are ultimate analysis, thermal quantities among which caloric value (HHV), fibers distribution and equilibrium moisture content. The obtained results confirm that torrefaction is a viable thermal pre-treatment to upgrade biomass and looks promising for a new generation of biomass derived fuel. As main topic, this research focuses on evaluating the role of the Mass Yield (MY) as synthetic parameter of the process. In particular it is made evidence that the properties of the torrefied samples can be led back to the MY without making reference to their thermal-time pathway. As example, considering two samples presenting the same MY of 80%, the former torrefied at 280 °C for 76 min, the second at 310 °C for 17 min, it is verified that the values of the aforementioned properties are very similar for both the samples even if submitted to different thermal pathways. The potentiality of the MY has been enhanced by extending this investigation to additional species of biomass pertaining to two groups, woody biomass and non-woody biomass. Inside the conditions ranges of the proposed experimental tests and for each of the indicated biomass type, an accurate linear correlation is obtained between MY and Energy Yield (EY) with a determination coefficient of ( R 2 ) 0.98 and 0.97 for woody and non-woody biomass respectively. Besides confirming the benefit of the torrefaction in enhancing the quality of the treated biomass, the use of the MY as synthetic parameter can be exploited to improve the torrefaction modeling schemes and to optimize the selection of the process working conditions. The proposed approach can therefore be useful to examine the relevant amount experimental data till now produced on this issue in view of exploiting and orienting their use towards real scale plant design.