In frozen dough applications a prefermentation period during the preparation of the dough is unavoidable and might also be important to obtain bread with a good texture. A major disadvantage of the prefermentation period is that it is associated with a rapid loss of the freeze resistance of the yeast cells. A major goal for the development of new baker's yeast strains for use in frozen dough applications is the availability of strains that maintain a better freeze resistance during the prefermentation period. We have isolated mutants that retain a better stress resistance during the initiation of fermentation. Some of these showed the same growth rate and fermentation capacity as the wild type cells. These mutants are called 'fil', for deficient infermentation induced loss of stress resistance. First we used laboratory strains and heat stress treatment, given shortly after the initiation of fermentation, as the selection protocol. The first two mutants isolated in this way were affected in the glucose-activation mechanism of the Ras-cAMP pathway. The fil1 mutant had a partially inactivating point mutation in CYR1, the gene encoding adenylate cyclase, while fil2 contained a nonsense mutation in GPR1. GPR1 encodes a member of the G-protein coupled receptor family which acts as a putative glucose receptor for activation of the Ras-cAMP pathway. In a next step we isolated fil mutants directly in industrial strains using repetitive freeze treatment of doughs as selection protocol. Surviving yeast strains were tested individually for maintenance of fermentation capacity after freeze treatment in laboratory conditions and also for the best performing strains in frozen doughs prepared with yeast cultivated on a pilot scale. The most promising mutant, AT25, displayed under all conditions a better maintenance of gassing power during freeze-storage. It was not affected in other commercially important properties and will now be characterised extensively at the biochemical and molecular level.