This paper presents a new fundamental static state estimation algorithm using weighted least square (WLS) estimation, which is based on singular value decomposition (SVD) rather than the normal equations. In addition, this paper also uses linear WLS estimation, which is quicker than non-linear. The SVD approach does not require the whole network system to be observable prior to estimation. It can provide a solution even if the system under consideration is partially observable. The simulation study was performed on the IEEE 14-bus test system. The simulation results, both linear and non-linear WLS, have shown that the SVD approach can provide a solution even when ill-conditioned occurred while the normal equation approach failed to give satisfactory results. In addition, the SVD approach can identify which parts of the network are unobservable islands. © 2003 Elsevier Science B.V. All rights reserved.