A pear breeding programme was developed at INRA Angers-Beaucouze from the 1960s and is still in progress. In the 1960s, the programme put emphasis on late maturing fruit with a long storage ability; from the 1980s, the programme was mostly devoted to fire blight resistance. A large amount of crosses and seedlings were scored along the years, giving rise to pear selections which were trialled for a future release. The seedlings from the first set (1960s) were grown without any greenhouse selection and only screened in the field for traditional quality traits. The seedlings from the other set (1980s) were grown after selection in the greenhouse for fire blight low susceptibility, and then transferred in the field for screening on the same traditional quality traits. The resultant selections from the 2 sets of crosses were grafted and scored in the greenhouse for fire blight resistance along a period of 30 years. It appears that the process of selection in the greenhouse is not leading to promising pear selections with a very good score for fire blight resistance. These results are discussed for proposing a strategy which will be efficient for screening new cultivars with a better score for fire blight resistance.