Responses of sugarcane genotypes across environments in north Queensland for sugar yield and its components cane yield and sugar content, were examined using pattern analysis methods. The aim was to provide information which could be used in designing efficient selection methods for sugarcane breeding programs. Of particular interest was the comparison of variation in relative performance of genotypes across sites and crop-years. Results from two series of sugarcane trials in north Queensland are presented. In the first series, 65 genotypes were evaluated at four sites and across plant, first ratoon and second ratoon crops. In the second, 52 genotypes were evaluated across four sites and across plant and first ratoon crops. Crops and years were confounded in both series. Crop-years within sites were generally more similar for genotypic responses than environments from different sites. Within most sites the relative ranking of groups of genotypes derived from cluster analysis was similar across different crop-years. However, there were some substantial differences in ranking of the groups between some sites and selection would vary if different subsets of sites were considered. There was little variation among groups in general response to ratoon crop-years. The results indicate that testing across sites is of importance. However, the similarity in information obtained across crop-years within most sites suggests there may be limited gain in testing across multiple crops or years within a particular site.