Summary: We show that empirical likelihood is justified as a method of inference for a class of models much larger than the class of linear models considered by

*A. Owen* [Ann. Stat. 19, No. 4, 1725-1747 (1991;

Zbl 0799.62048)]. In particular, we show how empirical likelihood may be used with generalized linear models. Quasi-likelihood and extended quasi- likelihood are used to derive the necessary estimating functions, but the method can be applied similarly using other sources. We consider separately those models in which the dispersion parameter is fixed and known, those in which it is fixed but unknown, and those in which it is itself modeled linearly via a link function.