Between late May and early October of each year, marine stratus encroaches upon interior regions of the San Francisco Bay Area. Ceilings below 3000 ft impede landings at San Francisco International Airport (SFO) in that the parallel runways are sufficiently close together that aircraft must maintain visuals on approach to utilize the maximum 60 per hour landing rate. If visuals are not met the landing rate drops to 30 per hour. This results in a high number of arrival delays during the summer stratus season that can produce major impacts on the National Air Space system. The Marine Stratus Initiative, a 8 year research effort funded by NASA from 1995-2003, was an effort to develop a decision support system to aid in the prediction of when stratus would clear sufficiently to allow unrestricted landings at SFO. The ultimate goal was to avoid issuance of ground delays when there was high confidence of clearing prior to landing rates exceeding a 30 rate, and thus improve the efficiency of the National Air Space. The decision support system (Figure 1) issues its forecast in both deterministic (consensus of clearing times using up to four algorithms) and probabilistic terms, i.e. a calibrated probability of clearing associated with arrival times beginning at 17Z, the historical onset of arrival rates exceeding 30. Probabilities are provided hourly out through 20Z for each bi-hourly forecast cycle beginning at 09Z each day. This system was declared operational by the FAA in 2004 and subsequently transferred to and funded by the National Weather Service. The program has been running in an operational mode for the period from June 2004 through the summer of 2007. This study attempts to quantify whether implementation of this decision support system has lead to a meaningful reduction in arrival delays at SFO during the period from 2004-2007. To verify the link between the decision support system and arrival delays, the analysis first filtered days where flights operations were only related to marine stratus intrusions. It then correlated percent of on-time arrivals, minutes of delay per arrival, and number of hourly delays with forecasted hourly probabilities for clearing at SFO. The decision to issue a ground delay program for SFO is made each stratus day by approximately 1315Z by a consensus of human inputs including the FAA Command Center, the Traffic Management Unit at the FAA Oakland ARTCC, the collocated NWS Center Weather Service Unit, the NWS Terminal Aerodrome Forecast, and United Airlines. The final consensus forecasts may or may not agree with the deterministic and/or probabilistic forecasts issued by the decision support tool. Thus human factors must be accounted for in determining the utility of the decision support system. In addition, recent trends for an increasing number of arrivals as early as 15Z each day has impacted the utility of the decision support system based on the original design. These impacts will be discussed along with what modifications to the system are needed to improve the systems utility under current airport operating conditions. Ivaldi, C., D.A. Clark, and D. Reynolds: 2006: Upgrade and technology transfer of the San Francisco Marine Stratus Forecast System to the National Weather Service. 12th Conference on Aviation, Range, and Aerospace Meteorology, Atlanta, GA. Amer Meteor. Soc.