Mill Creek in Cleveland, Ohio, receives discharges from combined-sewer overflows (CSOs) and other sanitary-sewage inputs. These discharges affect the water quality of the creek and that of its receiving stream, the Cuyahoga River. In an effort to mitigate this problem, the Northeast Ohio Regional Sewer District implemented a project to eliminate or control (by reducing the number of overflows) all of the CSOs in the Mill Creek watershed. This study focused on monitoring the microbiological water quality of the creek before and during sewage-collection system modifications. Routine samples were collected semimonthly from August 2001 through September 2004 at a site near a U.S. Geological Survey stream gage near the mouth of Mill Creek. In addition, event samples were collected September 19 and 22, 2003, when rainfall accumulations were 0.5 inches (in.) or greater. Concentrations of Escherichia coli (E. coli) were determined and instantaneous discharges were calculated. Streamflow and water-quality characteristics were measured at the time of sampling, and precipitation data measured at a nearby precipitation gage were obtained from the National Oceanic and Atmospheric Administration. Concentrations of E. coli were greater than Ohio's single-sample maximum for primary-contact recreation (298 colony-forming units per 100 milliliters (CFU/100 mL)) in 84 percent of the routine samples collected. In all but one routine sample E. coli concentrations in samples collected when instantaneous streamflows were greater than 20 cubic feet per second (ft3/s) were greater than Ohio's single-sample maximum. When precipitation occurred in the 24-hour period before routine sample collection, concentrations were greater than the maximum in 89 percent of the samples as compared to 73 percent when rainfall was absent during the 24 hours prior to routine sample collection. Before modifications to the sewage-collection system in the watershed began, E. coli concentrations in Mill Creek ranged from 220 to 29,000 CFU/100 mL. After major modifications, E. coli concentrations ranged from 110 to 80,000 CFU/100 mL. The percentage of sample E. coli concentrations in the former group greater than Ohio's single-sample maximum was 88 percent, whereas 85 percent of sample concentrations was greater than the maximum after major modifications occurred. Instantaneous discharges of E. coli were calculated for each of the modification periods. No statistically significant difference was observed between the median instantaneous discharges of E. coli for the premodification and minor-modification periods (5.1 ? 106 and 3.6 ? 106 CFU per second, respectively). During rainfall events in September 2003, samples were collected every 15 to 30 minutes. E. coli concentrations in all of these samples (n = 34) were greater than Ohio's single-sample maximum for primary-contact recreation. On September 19, total accumulated rainfall was 1.7 in., and streamflow reached a peak of 1,040 ft3/s. Sample collection started after 0.8 in. of precipitation had fallen and continued throughout the remainder of the storm. For these samples, E. coli concentrations ranged from 32,000 to 140,000 CFU/100 mL. On September 22, total accumulated rainfall was 0.5 in., and streamflow reached a peak of 497 ft3/s. Sample collection began before the start of the rain and continued throughout the storm. E. coli concentrations ranged from 450 to 260,000 CFU/100 mL.