The focus of this paper is a re-sampling technique, namely the bootstrap, for assessing the reliability of combining data from a large number of individuals in a conjoint analysis procedure. The issue of aggregation of preferences is a crucial one in large-scaled surveys where it is often used to simplify the analyses and obtain results in a more concise and therefore manageable form. The bootstrapping technique is applied to two large sets of data collected by an insurance firm. Results strengthen the relevance of this technique. The computer program, using the SAS package, is presented.