Human reliability in computer programming can be improved by reducing human errors. The traditional approaches for error reduction in industry are not applicable to minimizing errors in computer programming. Proposes a model for error reduction in software prior to its final release. The model consists of two modules, an error detection module, and an error correction module. A computational procedure is outlined for determining the optimal number of detection and correction stages prior to the final release of the software.