This paper develops a signal optimization algorithm that aims to equalize queue growth rates across links in oversaturated urban roadway networks and thus postpones queue spillbacks that form at the localized sections of networks. The performance of this algorithm is evaluated by simulating traffic conditions with optimized signal settings on an idealized 3 by 3 roadway network under various oversaturated demand scenarios. The simulation experiments show that the algorithm can delay queue spillbacks by distributing queues over upstream links that would otherwise be underused. The findings from the experiments also show that the signal settings optimized by the queue growth equalization (QGE) algorithm outperform those optimized using the conventional signal optimization software, TRANSYT-7F, for all the performance measures examined in this paper, i.e., compared with TRANSYT-7F, the QGE results in higher outflows, higher vehicle miles traveled, shorter delays, less sensitivity to various demand scenarios, and delayed queue spillbacks. In addition, the algorithm is computationally light to provide a promising groundwork for large-scale signal optimization.