This paper faces the problem of reducing costs of genotyping and recording traits not routinely measured in QTL detection programs for dairy sheep outbred populations. With this aim, starting from the actual population structure of the Sarda breed selection nucleus, QTL detection power of daughter design, granddaughter design, and alternative strategies as selective genotyping or selective phenotyping was calculated. Results were discussed focusing on global costs and practical implications of applying QTL detection designs in dairy sheep outbred populations.