Summary It has been suggested that disconnected data pose problems in variance component estimation and that the dlscarding of certain sections of a data set is a possible solution. In contrast, it is shown in this paper that estimators of variance components can be derived from disconnected data by applying Henderson's Method III without modification. The entire data set is utilised, and the subsequent estimators are generally more efficient than those obtained from connected subsets or combinations of them. Similar conclusions holdfor the MINQUE method of estimation. A simple example which illustrates the results of the paper is given. Recently Schaeffer (1975) discussed the estimation of variance components from disconnected data. He examined a numerical example using the MINQUE method and obtained different estimates of the variance components when different connected subsets of data were used. It was suggested that perhaps only the largest connected subset of data should be used and the remaining data discarded. Such data manipulation is unnecessary if not wasteful since the MINQUE method is not dependent on whether or not data are connected. Clearly estimates can be obtained by the MINQUE method from all the disconnected data and from any individual connected subsets that contain sufficient data. Henderson's Method III (HMIII), which yields explicit expressions, is applied in order to determine the effects that disconnectedness may have on the estimation of variance components.