The conventional computing techniques are good at performing fast arithmetic and computations as per programme's instruction set in the program. But conventional computing is not so good to interact with noisy data or data from the environment, massive parallelism, fault tolerance, and adapting to circumstances. In same scenario the classification of diseases are discussed and addressed in this paper. Here the noisy data with high level of parallelism is considered and we always need to adapt it as per circumstances and situations based on applications. So it becomes difficult to adapt conventional computing for classification of diseases. On other hand the neural network (NN) systems usage for formulation of algorithm and provide the solutions is efficient and faster. Also NNs has capabilities to extract meaningful information from highly complicated data, and it can be used to predict the patterns or to analyse trend in complex scenario which can't picked by humans or by other conventional computing. When NN get fully trained it will be considered as expert for particular information that it has been given to analyse. In this paper we have carried a comprehensive assessment and performance overview of Neural Networks for Classification of Diseases. This can be used in densely populated countries or areas having lack of medical facilities are very difficult to classify the disease of patients at the earliest and prescribe the proper treatment on the basis of correct diagnosis. In practice, classification of diseases is a tedious and cumbersome task in terms of computation. On the basis of capabilities of Neural network, in this paper wehave addressed overviewed and studied comprehensively why, what, when, where, who, and how one can classify disease effectively using ANN?