The effectiveness of variable “advisory” speed limit (VASL) systems in congested urban work zones was investigated. Except for one publication, all previous studies focused on regulatory speed limit systems. This study used a comprehensive set of performance measures to investigate VASL effectiveness. Four congested work zones with lanes reduced from four to three lanes on Interstate 270 in St. Louis, Missouri, were selected for empirical and simulation analysis. The empirical analysis showed that VASL were effective in slowing down drivers gradually as they approached the work zone, thus reducing any sudden speed changes. Simulation analysis showed that operationally, the use of VASL resulted in: a 39% to 53% decrease in average queue length, a 7% to 11% reduction in throughput, a 4% to 8% increase in travel time. The use of VASL achieved a decrease in the standard deviation of speeds at the taper and 1-mile upstream of the work zone. The maximum speed differences also decreased up to 10 mph with VASL. A new VASL algorithm investigated in the study showed significant improvements in mobility and safety performance compared to the original field algorithm. The new algorithm made promising improvements in safety—a 30% reduction in rear-end conflicts and a 20% reduction in lane changing conflicts as compared to the without VASL condition. The new VASL algorithm also decreased the queue length, further improving the overall safety in congested work zones. Thus, it is possible to design a VASL system that improves traffic and safety in congested work zones.