Chemically functionalized or coated sensors are by far the most employed solution in gas sensing. However, their poor long term stability represents a concern in applications dealing with hazardous gases. Uncoated sensors are durable but their selectivity is poor or non-existent. In this study, multi-parametric discrimination is used as an alternative to selectivity for uncoated capacitive micromachined ultrasonic transducers (CMUTs). This paper shows how measuring simultaneously the attenuation coefficient and the time of flight under different nitrogen mixtures allows to identify hydrogen, carbon dioxide and methane from each other and determine their concentration along with identification of temperature and humidity drifts. Theoretical comparison and specific signal processing to deal with the issue of multiple reflections are also presented. Some potential applications are monitoring of refueling stations, vehicles and nuclear waste storage facilities.