An analysis of the risk of progression towards leukemia, carcinoma and myelofibrosis was performed in 93 patients treated by 32P alone (PVSG protocols) since 1970-1979, 395 patients over the age of 65 years treated by 32P with or without maintenance therapy using hydroxyurea (French protocol) since 1980-1994, and 202 patients under the age of 65 treated by either hydroxyurea or pipobroman since 1980. The risk of leukemia, or myelodysplasia, or lymphoma in the 32P-treated patients was 10% at the 10th year, but increase after that time to reach a value of about 30% at the 20th year, in the surviving case. This risk was not dose-related. Despite a marked reduction of the cumulative 32P dose in the patients maintained by hydroxyurea, the actuarial risk was 19% at the 10th year. In the patients treated exclusively by non radio-mimetic agents (hydroxyurea or pipobroman) a risk of 10% at the 10th year was observed. The risk of carcinoma (excluding skin cancers) was about 15% at the 10th year in the 32P-treated cases, a value similar to that generally reported by the French statistics. There was no prevalence of digestive carcinomas. In contrast, the patients receiving 32P and hydroxyurea as maintenance had an excess risk: 29% at the 10th year. In the relatively young cases treated by non radio-mimetic agents, the risk was similar in both arms: 9% at the 10th year, similar to the expected incidence at this age. The risk of myelofibrosis with myeloid metaplasia was still relatively low at the 10th year, about 15% in all arms, but increased towards a value higher than 30% in the patients surviving at the 20th year. At the present time, but in only a few cases with long-term following, no myelo-fibrosis with splenic metaplasia has been observed in the pipobroman-treated cases. The present results, which need to be confirmed (the present analysis has been done in spring 95) suggest that:-the use of non radio-mimetic agents does not protect against leukemic transformation, which may be a consequence of the disease; rather than of the treatment,-maintenance therapy after initial use of 32P increases the risk of both leukemia and carcinoma,-and hydroxyurea does not delay the risk of developing myelo-fibrosis, in comparison with 32P alone.