The process of picking products from their storage locations to fill customer orders is a very critical and labour intensive logistic activity, especially in presence of many and small orders. The storage strategy is one of the main factor impacting the order picking system performance. Object of this paper is the development and application of an original and systematic procedure for the minimization of the variable travelling cost in an OPS with tri-lateral stackers. This procedure is based on the introduction of similarity indices, as measures of correlation between products, and on the application of hierarchical clustering algorithms. The authors propose different storage allocation/positioning rules supported by isotime mapping of the storage area. A case study is illustrated by the discussion of the results obtained by an experimental analysis.