This new study complements the mortality study of a cohort of veterans who were present at French nuclear test sites in the Pacific between 1966 and 1996 and who underwent dosimetric monitoring. Its objective is to assess the incidence of long-term illnesses (ALD) among cohort members still alive in 2003, over the period 2003–2008. The study population consists of 18,717 veterans. All ALDs occurring between 2003 and 2008 and recorded in the inter-insurance system database (Sniiram) were considered. The incidence of ALDs by type was compared to that of the general population using standardized incidence ratios. To test for the effect of non-zero dosimetry, Poisson models were used. From 2003 to 2008, 4,887 new ALDs were treated in 3,584 men. Tumors are the leading cause of ALDs (30%). Chronic arterial diseases and cancers of the prostate and mouth are overrepresented compared to the general population, but are not associated with dosimetry. In the cohort, two conditions with a low number of ALD cases—multiple sclerosis and peritoneal cancer—show significant excesses in the group with non-zero dosimetry. While the mortality study had shown an excess of malignant hematological diseases among veterans with non-zero dosimetry, this new study—limited to a short period and relatively distant from the exposure—does not reveal an excess of new cases of ALD potentially linked to radiation among veterans. (R.A.)