Data are available on a 30-year follow-up of miners, ex-miners and miners from the Rhondda Fach mining valley in south Wales. The data are in the form of survival times/times of loss to follow-up and a number of covariates. The aim of the analysis is to find out if the mortality of the miners and ex-miners is different from that of the non-miners. In addition to mortality from all causes, mortality from specific causes is to be analysed to find out if there are any causes (other than coalworkers' pneumoconiosis) for which miners and ex-miners have an increased risk. Where possible allowance is to be made for the effects of the covariates like age, weight and body type. A number of methods for regression analysis of censored survival data are used and their results compared. These methods include Standardised Mortality Ratios (SMRs), log-linear models for grouped data, models for fixed period survival, proportional hazards models, and accelerated time models. Computer software is generally available for implementing most of the methods, and where it is not, it can easily be developed using standard libraries of subroutines. However, the more comprehensive parametric models are not easy to fit because of the difficulty in computing the gamma, beta and related functions. The results given by the various methods are generally similar and imply that, if only a few covariates are being considered, simple models using grouped data are quite adequate. The results show increased mortality among miners and ex-miners. The increase is accounted for by increased mortality from pneumoconiosis, bronchitis, and other non-cancerous respiratory diseases; and in the case of miners and ex-miners with simple or no pneumoconiosis from accidents. The Rees/Eysenck Body Type and weight have significant effects on mortality, particularly mortality from non-cancerous respiratory diseases. Apart from the results provided by more sophisticated analysis, no additional information seems to have been provided by the 30-year follow-up, that was not available from the 20-year follow-up.