Abstract Accurate prediction of damage to petroleum production infrastructure from hurricane winds, waves, and storm surges has numerous applications to facility design, production planning, and the financial markets, among others. Here we describe a system for the estimation of damage, repair time, and shut in production using a numerical hurricane model and damage models for various infrastructure components. The system was designed and pilot-tested using data from Hurricane Ivan (2004) and prior storms, and validated operationally in forecast mode during the 2005 hurricane season. The petroleum extraction and distribution systems in the Gulf of Mexico as of 1 May 2005 were stored in a geographic information system. Damage to that infrastructure, consisting of over 50,000 elements, was computed using a three-dimensional physics hurricane model driven by forecast track and intensity data. The resulting damage statistics were then used to derive shut in production predictions for the Katrina and Rita landfalls, the results of which were posted on a web site in real time. The performance of the real time forecast system has been compared to reported statistics and found to accurately predict shut in production, especially using 24 hour forecasts through immediate post storm meteorological data. The validated shut in production prediction model was then applied using the 155 years of existing historical data to produce estimates of the magnitude and frequency with which significant production disruption would have been expected from the Gulf of Mexico facilities, given the present distribution of platforms and on-shore facilities. These frequencies indicate that significant disruptions of production on the order of magnitude seen in 2005 would be relatively common occurrences, given the performance of the infrastructure in recent events and historical hurricane activity. This implies that both design criteria and management of the petroleum production and distribution infrastructure should be revised.