Output gap estimates are constructed for India using unobserved components model (UCM) approach on the lines of Watson (1986) and Kuttner (1994). Results from UCMs are not found to be any less sensitive to data revisions when compared to those from the Hodrick-Prescott filter. This, however, could be because of lack of sufficient 'revised-data' on which the sensitivity of the results can be tested. Based on standard deviation of change in potential output to data revisions and its 'economic' content, the UCM using trimmed mean as the numeraire for inflation comes forth as the best choice. Alternative estimates of "core" inflation, included as a state variable in one of the UCMs, are also provided