This paper proposes a conceptual clustering approachfor RNA sequences using codons. It is shown thatemploying the codons (codon usage model) in the conceptualclustering of RNA sequences has high efficiency and robustnesscompared to conventional clustering methods. In cases wherethere are hidden structural patterns, homology search algorithmsare inefficient in locating similar sequences and as a result arenot reliable in the task of biological sequence clustering. As isshown by empirical results in this paper, conceptual clusteringusing the codons is able to discover similar sequences in adatabase of sequences with hidden structural homologues. Thecodon usage and cohesiveness model introduced in this papercan be efficiently employed in clustering biological sequence datawhere conventional homology search algorithms fail.