Genes in higher eukaryotes extend through many thousands of base pairs in the chromosome, but the regions of those genes that code for proteins usually occupy only a small percentage of the sequence. Identifying these coding regions is of vital importance to the understanding of genetics, but traditional laboratory techniques for finding genes are costly and time-consuming. A computational gene finder must combine evidence from various sources reflecting the content of a sequence as well as specific biological signals that appear as statistically recognizable patterns. This talk will describe how decision tree classifiers, Markov chains, and dynamic programming are integrated in the MORGAN gene-finding system. Current performance plus promising avenues for future improvements will be highlighted. This is joint work with Arther Delcher, Kenneth Fasman, and John Henderson.