In this paper we generalize the fundamental methods of process data reconciliation developed by Crowe (AIChE J. 1986, 32, 616-623), to allow for the presence of censored concentration measurements in the bilinear mass balance equations. The resulting statistical analysis leads to the maximization of a general likelihood function subject to linear constraints. The difference with respect to both Crowe＇s analysis and a previous algorithm that considered only linear mass balances (Dovi et al. Chem. Eng. Sci. 1997, 52 (17), 3047-3050) is considered, and an example is examined for illustration purposes.