A robust analysis of comparative genomic microarray data is critical for meaningful genomic comparison studies. In this paper, we compare our method (implemented in a new software tool, GENCOM, freely available at http://www.ifr.ac.uk/safety/gencom ) with three commonly used analysis methods: GACK (freely available at http://falkow.stanford.edu ), an empirical cut-off value of twofold difference between the fluorescence intensities after LOWESS normalization or after AVERAGE normalization in which the fluorescence intensity is divided by the average fluorescence intensity of the entire data set. Each method was tested using data sets from real experiments with prior knowledge of conserved and divergent genes. GENCOM and GACK were superior when a high proportion of genes were divergent. GENCOM was the most suitable method for the data set in which the relationship between the fluorescence intensities was not linear. GENCOM has proved robust in an analysis of all the data sets tested.