The spatial and temporal network dynamics induced by variable message sign (VMS) information under nonrecurrent congestion are investigated. Computational experiments were performed to analyze the role of incident attributes (number, location, and timing), information characteristics (delay, update frequency, compliance rates), and alternative operational strategies. The network performance of VMSs was compared with those of in-vehicle devices (IVDs). The findings indicate that VMS performance varies dynamically and depends on the interactions between information lag, diversion rate (overconcentration) onto alternative paths, efficiency of reported paths, time-varying interactions between vehicles with VMS information and vehicles without VMS information, residual capacity on alternate paths, compliance rates, and spatial incident characteristics. The VMS strategies that account for information quality, compliance, and overconcentration are likely to outperform purely travel time-based strategies. The findings have important implications for the design, operation, and evaluation of VMS and IVD systems.