Block cipher cryptanalysis is a very difficult problem for which, so far, no automated solution has been developed. Common approaches, such as linear or differential cryptanalysis, take advantage of the discovered weaknesses exhibited by a cipher. These weaknesses, however, are only discovered through painstaking, intense scrutiny by professional cryptanalysts. Most of the exploits that have been found to break modern ciphers are based on the observation that those ciphers do not produce truly random output. In this paper, we extend the work of Hernandez, et al, in which genetic algorithms are used to solve the problem of determining whether a given cipher produces random output. We show that carefully tailored genetic algorithms are capable of finding efficient distinguishers for ciphers much faster than has previously been reported.