A methodology is presented for estimating and predicting the total annual traffic congestion attributable to recurrent and nonrecurrent congestion. The methodology is applicable to freeways, conventional highways, and urban streets. The methodology is sensitive to improvements in facility surveillance and freeway service patrol strategies. Recurrent congestion is estimated by using Highway Capacity Manual speed-flow curves and data on facility demands, free-flow speeds, and capacities. Nonrecurrent congestion is estimated in terms of annual vehicle hours of delay caused by weather, work zones, incidents, and so forth. The methodology employs incident probability trees, incident duration (sensitive to surveillance and response times), and estimates of remaining capacity during incidents to estimate incident delay. Weather- and work-zone-related delays are estimated on the basis of frequency of occurrence and estimates of capacity reductions during periods of bad weather and work-zone activity. The methodology has been implemented in a spreadsheet.