Research was conducted to better understand the breakdown phenomenon and look for ways of strategically increasing capacity and travel speed by reducing breakdowns on a section of freeway in Mississauga, Ontario, that is subject to ramp metering. The major task was to quantify the probability of breakdown as an increasing function of volume at the critical location. Data from 71 peak periods during which breakdown occurred at the critical merge were used to estimate the function, which was then used in a computer simulation of fixed- and variable-rate metering to demonstrate the potential benefits of using the probability-of-breakdown concept as a basis for ramp metering. The results indicate, for example, that variable-rate metering that complements a constant merge flow limit of 2,320 veh/h per lane could increase peak-hour throughput from about 6,460 to 6,600 veh/h at this location. They also indicate that 2,500 veh/h might be allowed, provided stringent metering for quick recovery to free flow is feasible.