As part of the U.S. Geological Survey's (USGS) National Water-Quality Assessment (NAWQA) Program, urban indicator sites were monitored to (1) characterize the stream quality from drainage basins with predominantly residential and commercial land use, and (2) determine which selected natural and anthropogenic factors affect stream quality. A total of 869 water samples were collected from 37 urban streams during 1995-2003 and were analyzed for 87 volatile organic compounds (VOCs). The occurrence of VOCs in urban streams is described in this report for (1) all samples as a single dataset, (2) all samples grouped by streamflow pentiles, and (3) all samples grouped by warmer (April through September) and cooler (October through March) months by the detection frequency and (or) concentration of (a) any VOC, (b) VOC groups, and (c) individual compounds. An assessment level of 0.02 microgram per liter (ug/L) was used to compute the detection frequencies and concentrations of VOCs. Concentrations of VOCs were compared to (1) U.S. Environmental Protection Agency's (USEPA) drinking-water Maximum Contaminant Levels (MCLs) or Drinking Water Advisories, (2) Health-Based Screening Levels (HBSLs) developed by the USGS in collaboration with the USEPA and other agencies, and (3) USEPA and Canadian aquatic-life criteria. One or more VOCs were detected in 97.1 percent of 869 samples, and one or more VOCs were detected frequently (greater than 80 percent) at all sites. The median total VOC concentration for all samples was 0.57 ug/L, and total VOC concentrations in a single sample ranged from not detected to 698 ug/L. About 85 percent of the samples contained two or more VOCs, and about one-half contained five or more VOCs. The gasoline hydrocarbons were the most frequently occurring VOC group followed by solvents, trihalomethanes (THMs), gasoline oxygenates, organic synthesis compounds, fumigants, and refrigerants. Concentration ranges for most VOC groups were distributed over at least two orders of magnitude. Fifty-seven of the 87 VOCs analyzed were detected in at least one sample at an assessment level of 0.02 ug/L. More than one-half of the 30 VOCs not detected in samples were organic synthesis compounds. Fifteen compounds had detection frequencies greater than or equal to 10 percent. With the exception of toluene and chloroform, the median concentration of each VOC for all samples was less than the assessment level. Furthermore, the median concentrations of detections for the 15 most frequently occurring VOCs ranged from 0.03 to 3.9 ug/L, and typically were less than or equal to 0.10 ug/L. The 869 samples from the 37 sites were stratified into five streamflow pentiles (less than 20, 20-less than 40, 40-less than 60, 60-less than 80, and greater than or equal to 80 percent of estimated long-term streamflow statistics) for comparison of the occurrence of VOCs. The detection frequency of one or more VOCs by streamflow pentile varied only slightly from 96.7 to 97.7 percent. The median total VOC concentrations in samples for the five streamflow pentiles ranged from 0.39 to 1.0 ug/L. Two or more VOCs were present in more than 80 percent of samples in each of the five pentiles. The gasoline hydrocarbons, solvents, THMs, and gasoline oxygenates occurred frequently (greater than 30 percent) in all streamflow pentiles, in contrast to the organic synthesis compounds, fumigants, and refrigerants that occurred less frequently in urban streams under all streamflow conditions. The median total VOC concentrations for gasoline hydrocarbons, solvents, gasoline oxygenates, and organic synthesis compounds generally increased as streamflow increased. In contrast, the median total VOC concentrations for THMs and fumigants generally decreased as streamflow increased. The median total VOC concentrations for refrigerants showed no pattern as streamflow increased. Because differences between VOC occurrence and streamflow pentiles were small for most compariso