Verification and validation of AI systems is an important problem. In this article, we present a methodology for partial verification and validation of production system applications, in particular, OPS5-based applications. This methodology has two main parts in it. The first part, termed Domain-Independent Verification, involves finding certain properties and relations of productions by compile-time analysis. The second part, termed Systematic Testing, consists of generating the data-dependency graph, finding the set of distinct data-paths, and generating test data for each distinct data-path. We discuss how these two parts of the methodology contribute towards the partial verification and validation of OPS5-based applications. © 1991 IEEE.