Most of research efforts on the analysis of steady-state simulation focus on the estimation of the mean. However, analysts may be interested in other types of measures, (e.g. measures of risk) such as variance and quantiles. In this paper, we present an empirical comparison of the multiple replications, non-overlapping batches and spaced-batches methodologies to estimate mean, variance and 90%-quantile of an M/M/1 system. These methods are compared with respect to the empirical coverage of 90% confidence intervals through multiple experiments. The results show that spaced-batches exhibit poor coverage of the mean compared with replications and non-overlapping batches. The three methods perform similarly for estimating the variance and the 90%-quantile. However, estimating the variance requires more observations to obtain the expected coverage than the estimation of the mean and the quantile. © 2013 AIP Publishing LLC.