This paper presents an effort to develop a simulation testbed for the evaluation of ramp metering algorithms using CORSIM's Run-Time Extension (RTE), which provides exported functions and shared structures to communicate between the CORSIM microsimulator and the algorithms. A simulation network with 12 metered on-ramps for an 11-mile stretch of the I-95 corridor in Miami-Dade County, Florida was used to compare the results from three alternatives: fixed-time ramp metering control, Seattle fuzzy logic ramp-metering control, and no ramp metering. Results show that ramp meters are generally beneficial and that the Seattle fuzzy logic algorithm with only globally configured parameters outperforms a well-calibrated fixed-time control strategy especially under congested conditions. The simulation testbed developed in this research also makes possible the evaluation of additional ramp metering algorithms in a widely used simulation model in the United States.