For a widely varied wind climate in India, a single type of Extreme Value Distribution (EVD) is not suitable to fit wind speed data for all geographical locations and climatic conditions. This may lead to the inappropriate estimation of the design wind speed (Vd). In this study, a new approach, combining the merits of both the block maxima and the peaks over threshold approaches, has been used to determine the Vd. The hourly mean wind speed data of three stations, viz., Trivandrum, Bombay and Calcutta have been classified into block of months of the years 1973−2005, 1969−2007, and 1969−1994 respectively. The extreme peaks over the appropriate threshold value have been fitted into the generalized EVD (type-I, II and II). The results indicate that Fréchet distribution is the most suitable distribution for all three stations. March has the highest Vd for Trivandrum, whereas, for both Bombay and Calcutta it is September. Bombay shows highest overestimation of the Vd as given in Indian code of standard, whereas, for Trivandrum, and Calcutta these deviations have been found to be comparatively less. The five plotting position formulae have also been compared based on two goodness of fit tests, viz., R2, and RMSE.