species. In the marine environment, these campaigns must focus on a large enough number of edible marine produce samples from various trophic levels (flat fish, crustaceans, molluscs, etc.); 3. Given the major diversity of tritiated organic molecules, caution is needed in drawing conclusions and making extrapolations and the chemical speciation of discharges from potentially relevant sites needs to be characterised; 4. Given the still fragmentary nature of current knowledge on remanence and on tritium behaviour in sediment, targeted multidisciplinary studies with rigorous protocols need to be used to provide experimental verification of the hypotheses put forward in older studies, in particular regarding the possible influence of the activity of microorganisms in aquatic sediments when organic tritium is remobilised in aquatic animal organisms. In general, the scientific data regarding the conversion of tritiated water into organic tritium along the food chain should be enhanced. Reliable quantitative estimates are required. With respect to health effects, the group recommends the following (in priority order): 1. Up-to-date methods should be used to gather further knowledge on the effects of tritium, in order to enable comprehensive characterisation of the physico-chemical forms used (covering a concentration scale that would include industrial discharges) and the biological mechanisms at work, focusing not only on aspects relating to carcinogenesis, and looking at age at the time of exposure and at differences between accident- related and chronic exposure; 2. Knowledge of the effects of tritium exposure on embryos and foetuses should be improved. Further research in this area is vital; 3. Data regarding the potential induction of hereditary effects should be critically assessed and with great care. New approaches should be investigated, in light of the latest advances in biology; 4. The feasibility of epidemiological studies in French workers should be assessed, since it would be useful to gather data on tritium exposure and to process this data in a coordinated manner alongside other ongoing studies around the world; 5. A radiation weighting factor ( w R ) of 2 (instead of 1) should be used in individual risk assessment situations. No consensus was reached within the group as to the factor to be used in routine situations.