Recent modifications to the minimum energy efficiency requirements for residential water heaters have spurred an investigation into the variability in testing high-efficiency electric water heaters. While initial inter-laboratory comparisons showed excellent agreement between test results from different labs, subsequent inter-laboratory comparisons show differences between measured energy factors of up to 0.040. To determine the source of these differences, analyses of various parts of the test procedure are performed. For one case studied, the uncertainty in test results can be as high as +-0.028 if instrument accuracies reach the minimum level allowed in the test procedure. Other areas of the test procedure where variability is introduced are the optional use of pre-draws, the location of the lower tank temperature-measuring device, the use of insulation on tank fittings, and the use of a warm-up period before the simulated-use test commences. The implications of these issues on test results are provided.