The issue of an increased risk of leukaemia among people living near nuclear sites has been debated for more than 20 years. The story began in England, in November 1983. A local television station announced that a high number of leukaemia cases had occurred among the children living in Seascale, a village located three kilometres from the Sellafield nuclear fuel reprocessing plant. The existence of this excess was confirmed one year later (Black 1984). Since then, several localised clusters have been suggested in different countries, and tens of studies have been performed to assess the risk or to search for putative risk factors, mainly among young people (Laurier et al 2002). Even if radio-ecological studies showed that radiation exposure due to releases from the nuclear installations was small in regards to exposure due to natural sources (COMARE 1996, Rommens et al 2000), the existence of environmental risks due to nuclear installations remains of social concern. In the recent literature, studies may either focus on a specific site (Martignoni, 2003) or try to provide a general framework to locally suggested excesses (White-Konning et al 2004). The article from Yoshimoto et al (published on page 343 of this issue of the Journal of Radiological Protection) enters into the latter category. It brings the focus to the adult population, whereas almost all other studies focus on children only. It updates the analysis of the risk of leukaemia (excluding chronic lymphatic leukaemia) and Non-Hodgkin Lymphoma (NHL) mortality around nuclear power plants (NPPs) in Japan. In 1994, an increased risk of malignant lymphoma near one specific NPP was claimed in the Japanese press. A multi-site mortality study, considering all NPPs in Japan, was published one year later (Iwasaki et al 1995). Regarding the specific NPP, no excess of NHL risk was confirmed. Furthermore, the authors concluded that leukaemia and malignant lymphoma mortality in theNPPmunicipalities were not significantly different from that in the control areas. However, Hoffmann et al, analysing the same data, concluded with a global increase in leukaemia mortality risk in NPP municipalities compared to control areas, for all ages and for children below age 15 (not significant). They concluded that their results were ‘compatible with the hypothesis of a leukemogenic impact of nuclear power plants in Japan’ (Hoffmann et al 1996). Compared to the previous analysis, the new study from Yoshimoto et al presents several interesting features: • an increased number of NPPs (20 NPP municipalities), • an extended period of observation (26 years from 1972 to 1997), • a very large number of cases (more than 3100 leukaemia and 2700 NHL deaths), • an improved methodology of statistical analysis (Poisson regression, with different approaches to standardise or stratify on baseline risk variations). To reflect a putative risk associated with the operation of NPPs, the authors defined what they call a ‘surrogate score for potential radiation risk assessment’: 1 for residing in a NPP municipality at least 2 years after the start of commercial operation, 0 otherwise (that is residing in control municipalities without NPP, or in a NPP municipality before operation start plus two years).