Identifying genes within large regions of uncharacterized DNA is a difficult undertaking and is currently the focus of many research efforts. We describe a gene localization and modeling system, called GRAIL. GRAIL is a multiple sensor-neural network-based system. It localizes genes in anonymous DNA sequence by recognizing features related to protein-coding regions and the boundaries of coding regions, and then combines the recognized features using a neural network system. Localized coding regions are then "optimally" parsed into a gene model. Through years of extensive testing GRAIL consistently achieves about 90% of coding portions of test genes with a false positive rate of about 10% A number of genes for major genetic diseases have been located through the use of GRAIL, and over 1000 research laboratories worldwide use GRAIL on regular bases for localization of genes on their newly sequenced DNA.