Air sparging is an emerging method used to remediate saturated soils and groundwater that have been contaminated with volatile organic compounds (VOCs). During air sparging, air is injected into the subsurface below the lowest known depth of contamination. Due to buoyancy, the injected air will rise through the zone of contamination. Through a variety of mechanisms, including volatilization and biodegradation, the air will serve to remove or help degrade the contaminants. The contaminant-laden air will continue to rise towards the ground surface, eventually reaching the vadose zone, where the vapours are collected and treated using a soil vapour extraction (SVE) system. Air sparging performance and ultimately contaminant removal efficiency is highly dependent on the pattern and type of subsurface air flow. This paper presents the results of a laboratory experimental study which investigated the injected air flow pattern development within an aquifer simulation apparatus. The test apparatus consisted of a tank measuring 61 cm long by 25.4 cm wide by 38.1 cm high. The apparatus was equipped with one air injection well and two vapour extracton wells. Three different soils were used to simulate different aquifer conditions, including a sand, a fine gravel and a medium gravel. Experiments were performed with different injected air pressures combined with different vacuum and groundwater flow conditions. Experiments were also conducted by injecting air into simulated shallow aquifers with different thicknesses. The air flow patterns observed were found to depend significantly on the soil type, groundwater flow conditions and system controls, including injected air pressure, flow rate and applied vacuum. © Rapid Science Ltd. 1998