Data clustering consists of a group of procedures used to collect similar entries or data points within a set into clusters. No existing clustering technique considers entries sequentially in time. In some cases, it is desirable to generate clusters that represent a segment of a time-ordered data set. For these purposes, an order-specific clustering algorithm is proposed. The proposed algorithm employs representative load curves to describe the clusters it generates. The capabilities of the order-specific clustering algorithm are demonstrated on a case study using electricity demand data for the province of Ontario, Canada. Two different applications of the clustering algorithm on this data set are given to demonstrate the effect of error threshold values on the formation of clusters. An analysis of the error for each of these clustering applications is presented. (C) 2007 Elsevier Ltd. All rights reserved.