The information that groups of genes co-occur in several genomes provides a basis for further comparative genomic analysis. The task of finding such constellations, mostly referred to as gene clusters, has led to various models of increasing generality. A central feature to enhance the biological relevance of their definition when applied to real genomic data is to allow for slight differences in the gene content within a cluster, thus not only considering groups of exact equality. We contribute a model defining gene clusters as common intervals with errors and discuss different representations and the corresponding problems resulting for the search procedure.