Bacteria holding-time experiments of up to 62 h were performed on five surface-water samples from four urban stream sites in the vicinity of Atlanta, GA, USA that had relatively high densities of coliform bacteria (Escherichia coli densities were all well above the US Environmental Protection Agency criterion of 126 colonies (100 ml)^− 1 for recreational waters). Holding-time experiments were done for fecal coliform using the membrane filtration modified fecal coliform (mFC) agar method and for total coliform and E. coli using the Colilert^®-18 Quanti-Tray^® method. The precisions of these analytical methods were quantified. Precisions determined for fecal coliform indicated that the upper bound of the ideal range of counts could reasonably be extended upward and would improve precision. For the Colilert^®-18 method, analytical precisions were similar to the theoretical precisions for this method. Fecal and total coliform densities did not change significantly with holding times up to about 27 h. Limited information indicated that fecal coliform densities might be stable for holding times of up to 62 h, whereas total coliform densities might not be stable for holding times greater than about 27 h. E. coli densities were stable for holding times of up to 18 h—a shorter period than indicated from a previous studies. These results should be applicable to non-regulatory monitoring sampling designs for similar urban surface-water sample types.