The comparability of data from accredited laboratories is not always as good as customers would desire. A number of reasons for this fact are given. Clients of laboratories do not specify methodology, which may lead to the application of systematically different methods or to methodology with improper performance characteristics. Laboratories do not always validate methodology sufficiently which is, however, among others provoked by a lack of reference materials. It is argued that accreditation cannot enforce comparability on itself, but needs to be embedded in a “measurement infrastructure”. The quality systems which underly accreditation form a basic element. In addition, adequate methods, good laboratory performance studies and certified reference materials are required.