The aim of this work is to recognize handwritten characters of Indian language, Telugu. Single stage of classifying similar Telugu characters leads to low recognition rate. However similar characters of Telugu (Indian language) are recognized in two stages in the current work. Various preprocessing steps are carried out first to extract characters from the handwritten documents. The preprocessed characters are then utilized to extract features from them. These features are further used in the proposed two-stage classification. The misclassified characters from the first stage of classification are fed to the second classifier in the proposed method. The recognition rates obtained with the two stage system are better compared to the single stage classification system.