In this paper we present a statistical approach to natural gas consumption estimation of individual residential and small commercial customers. The approach is based on nonlinear regression principles. Parameters are estimated using mainly two real data sets - ordinary (approximately annual) meter readings of almost all customers and additional (approximately monthly) meter readings designed and operated within the frame of cooperation between the Institute of Computer Science of the Czech Academy of Sciences (ICS) and the West Bohemian Gas Distribution Company, a part of the RWE Group (WBG). The model was tested on various data sets. It has broad applicability in many areas of gas industry. (C) 2007 Published by Elsevier Ltd.