Hourly tide gauge data were analyzed for extreme sea level for three stations, Paradip, Vishakhapatnam, and Chennai, along the east coast of India. These stations are tidally dominant, with the magnitudes of tides exceeding those of surges, in general. During the period of analysis (1974--1988), two to four storm surge events were identified in a year, on average, with the maximum number occurring in Vishakhapatnam followed by Paradip and the least in Chennai. A statistical analysis involving the r-largest annual maximum method, the joint probability method (JPM) as well as the revised joint probability method (RJPM) was performed to estimate the return periods of extreme sea level. JPM gave an underestimate of levels for long return periods, which is improved in RJPM. The results of RJPM and r-largest methods are consistent and are within 95% confidence limits. The flexibility of the r-largest method is used to include two historical extreme sea level events in the analysis for Paradip, including the surges due to the super cyclone that occurred in October 1999. It is shown that the inclusion of these events showed an increase of about 79 cm for the estimated 100-year return period.