This paper describes a computer method that uses codon preference to help find protein coding regions in long DNA sequences. The method can distinguish between introns and exons and can help to detect sequencing errors.