Modern databases can be successfully used to develop computer-based identification systems. In a first case study, fish larvae were identified with an average of three easily obtained characters. In a second case study, 17 fish diseases out of 20 were diagnosed directly, using an average of six gross signs of a disease. A comparison of computer-based identification systems reveals that i) expert systems can be viewed as a very comfortable modern version of traditional identification keys, ii) numerical methods open the way to automatic identification but depend on expensive hard- and software, and iii) modern databases represent powerful, yet easy to use identification systems.