Type II restriction endonucleases cleave double stranded DNA molecules at sites characterized by one or more sets of nucleotide pairs sequences. These digestions are essential in such procedures as DNA cloning, DNA sequencing and restriction fragment length polymorphism (RFLP) analyses. A large number of enzymes with different sequence specificities are available. To date, most choices of restriction endonucleases have been made by trial and error. A computer program, REDI, has been developed that predicts the ability of a particular restriction enzyme to detect mutations. Characteristics of both the restriction endonuclease used and the DNA being cut are incorporated as variables in the program. The program was tested using mouse mitochondrial DNA (mtDNA) and bacteriophage lambda DNA because these have been sequenced and are well characterized. REDI was strongly correlated (rs = +0.862, n = 11, P less than 0.001) with mouse mtDNA RFLP detected by Ferris et al. [1] (Genetics, 105 (1983) 681-721). Even though predictions may be altered by a non-random association of nucleotides, which varies among DNA molecules, the predictions increase the probability of selecting the most efficient enzymes for use in the analysis of a particular DNA molecule.