In order to develop a detailed understanding of the dynamics of gas puffing (gas release as a series of distinct pulses) and more sustained degassing (steady plumes of gas) during persistent volcanic degassing, measurements of gas mass flux are required in the vicinity of the volcanic vent. Masaya volcano (Nicaragua), a persistently degassing system, is an ideal location for measuring the dynamics of releases of volcanic gas in the first few seconds of their propagation. We carried out two field experiments during February 2002 and March 2003, during which thermal infrared thermometers were targeted into the degassing vent at Masaya to record thermal variations related to variations in the at-vent gas emission over short (on the order of seconds) time scales. The thermometers recorded an oscillating signal as gas puffs passed through the field of view, detailing variations in the degassing process developing over time scales of seconds. These data were processed to extract puff frequencies, amplitudes, durations, emission velocities and volumes. These data showed that, over time periods of hours, the total gas flux was stable with little variation in the puffing frequency. However, between the 2002 and 2003 data set we noted an increase in mean plume temperature, puffing frequency, puff emission velocity and puff volume, as well as a decrease in mean puff duration. These changes were consistent with a thermal data-derived increase in emitted gas flux from 4.2 × 10 7 m 3 d − 1 to 6.4 × 10 7 m 3 d − 1 between the two campaigns. Turbulent gas puffs entrain surrounding air, and quantifying the magnitude of air entrainment, or dilution, represents a major challenge for the measurement of total volcanic gas emissions. Our observations of small gas puffs suggest that they behave as turbulent buoyant thermals, and we use equations for mass, momentum and buoyancy, coupled with the standard entrainment assumption for turbulent buoyant flows, to estimate the gas puff dilution. The theoretically calculated dilution of 0.09 and 0.24 between emission and detection yields total SO 2 mass fluxes of 209 t d − 1 and 864 t d − 1 for 2002 and 2003, respectively. This compares well with UV-spectrometer SO 2 fluxes of 470 and 680 t d − 1 for February 2002 and March 2003, respectively.