This article presents proposed changes in the methods used to estimate life expectancy by social class using the ONS Longitudinal Study (LS). The changes reviewed are: computational changes, including revised methods for age-specific mortality rates, more precise survival duration calculations and increased social class attribution through the inclusion of updated information -- extension of the criteria used to exclude LS members from the analysis, using information on presence at the 2001 Census -- the use of Health Authority deregistration data to approximate unrecorded emigration, together with information from the 2001 Census, to reduce a potential source of bias in mortality calculations. The impact on existing results is quantified and it is proposed that these methods are used in the updating of the published series.