Across Canada, thousands of abandoned gas wells leak gas to the surface. As a potential solution to this problem, we have done laboratory-scale experiments relevant to a field-tested novel process that injects a viscous oil-in-water emulsion to plug the formation near the wellbore. In an earlier study, we observed that, for unconsolidated cores, the process is effective and may withstand large pressure gradients over long periods of time; however, the emulsion penetration depth was limited to a small fraction of the cores' length. Now, we report results of subsequent experiments, our understanding of the emulsion blocking mechanism and criteria for controlling the distance into the formation to which an emulsion may penetrate. In these experiments, well-characterized emulsions were injected into an etched glass micromodel or a micromodel packed with well-sorted glass beads or sand grains to yield the desired permeability. Visualization experiments were done to observe the capture mechanisms of the emulsion droplets. We observed that, for a given pressure gradient, some droplets are too large to pass through a pore's throat (size exclusion), that other droplets attach to the pore's surface and coalesce with nearby droplets to accelerate the blockage process and that more viscous droplets are most effective in blocking pores. We found that the rate and extent of transfer of surfactant from the solution to the beads' surface must be important. In subsequent experiments, we flushed the micromodels with a surfactant solution to alter the wettability of the beads prior to injecting the emulsions. The results showed that the choice of surfactant, its concentration and the volume of its injected solution predictably affect the depth to which the oil-in-water emulsion may penetrate into a micromodel. In conclusion, this work characterizes the flow behaviour and breaking of an emulsion in a porous medium and examines parameters that may be adjusted to control the novel process.