In some stages of the breeding of crops, breeders must select the most promising lines from a very large number of sets of new varieties. Selection is conducted via direct comparison of new varieties that are tested within a single experimental unit, with control varieties systematically intercropped among them and replicated over a large number of plots. Researchers often do not realize that, by following some simple design rules, they might be susceptible of a precise and accurate statistical analysis. This paper discusses the subject with precision, and establishes rules of design and a statistical analysis technique appropriate for completely randomized experimental designs or randomized complete blocks with blocks of the same size.