Kervella et al (1991) made a bibliographic review of the experiments, covering papers published up to 1989. The feasibility of recurrent selection and the interest of applying it to pure line value breeding was demonstrated. Yet a number of different points still had to be optimized: intercrossing, early generation selection and multitrait selection. A complementary review which takes into account papers from 1990 to 1992 is presented here. Some older papers which did not appear in the first review are also analyzed. An evolution in the approaches, the methodologies and concerns of the authors is emphasized. Scheme efficiency is assessed; problems which remain to be solved and the limits of the experiments are discussed. recurrent selection / autogamous species / genetic gain / pure line value