It may be difficult to decide what distribution one's observations follow, but the form of the mean-variance relationship is often much easier to postulate; this is what makes quasi-likelihood useful. In"some of the cases where quasi-likelihood method 'is unable to provide reasonable results, a more general cla,ss of estimating functions, (Crowder, 1987) can be applied for better results. In this paper we describe the performance of these methods and explained it with some' examples.