We have recently shown that optimal growth temperature (T-opt) is one of the factors that influence genomic GC in prokaryotes. Our results have been disputed by Marashi and Ghalanbor, who claim that the correlations we show are not "robust" because the elimination of some points (arbitrarily chosen) leads, in some families, to variations in the correlation coefficients and/or significance of correlations. Here, we test whether the correlation between T-opt and genomic GC is robust by using two independent approaches: detection of possible outliers (using robust Mahalanobis distance) and usage of a non-parametric correlation coefficient that is not sensitive to the presence of outliers. The results presented here reinforce our previous proposal that Tort is correlated with genomic GC in prokaryotes. (c) 2005 Elsevier Inc. All rights reserved.