This paper presents a cost-benefit analysis of the Stockholm congestion charging system, based on the observed rather than on the model-forecasted data. The most important data sources are travel time and traffic flow measurements made in the year before the charges were introduced (during April 2005) and during the first spring with the charges (during April 2006, 4 months after the charges were introduced). Using matrix calibration, effects on the non-observed link flows and travel times are extrapolated, enabling us to calculate the social value of changes in travel times and travel costs. Impacts on traffic safety and emissions are calculated using standard Swedish CBA relationships. The system is shown to yield a significant social surplus, well enough to cover both investment and operating costs, provided that it is kept for a reasonable lifetime: investment and startup costs are "recovered" in terms of social benefits in around 4 years.