The prognosis of Hodgkins disease (HD) has improved during the last 30 years. This study was planned to analyse long-term survival of LID patients and to compare survival rates estimated from clinical trials and population-based data. Individual data were analysed on 2,755 adult HD patients entering randomised clinical trials of the British National Lymphoma Investigation BN LI) between 1970 and 1987, and 5,064 patients with HD incident 1978-1984 recorded in the UK population-based cancer registries participating in the EUROCARE study. Relative survival of Hodgkins disease patients allowing for mortality expected from general population rates was analysed by a proportional hazards regression model including covariates. Although relative mortality decreased with longer follow-up, it was still significantly positive at 9-10 years after diagnosis in both the clinical trials and the population-based data sets. Relative mortality was worse for late stage than for early stage patients even at 10-15 years after first treatment (BNLI data). Whereas 10-year relative survival was identical in trials and population-based patients at ages under 45 years (> 69%), it was much higher in BNLI older patients than in the population-based patients. In the older age group (65-74 years) the BNLI patients had 39% relative survival whilst for the population-based patients it was only 27%, Generalisation of clinical trials results to the general population must be done with caution, especially for older patients.