The multidisciplinary program "MOREST" aims to improve our understanding of causes of summer mortality in Crassostrea gigas juveniles in France and to reduce its impact on oyster production. As part of this program, four successive generations of families were bred between 2001 and 2005 in our research hatchery in La Tremblade. Survival and growth were recorded in the field in three sites along the French coast (Ronce, Rivière d'Auray and Baie des Veys). The first generation (44 full-sib families nested within 17 half-sib families) was tested during summer 2001 to assess to what extent genetic variability exists for spat survival. An important genetic basis was found (estimated narrow-sense heritability among sites = 0.83 ± 0.40). In contrast, lower genetic variation was observed for growth. From 2002 to 2005, response to selection for high or low survival was examined in several sets of outbred and inbred lines. Survival of the different progenies was determined to estimate realized heritability of survival and the repeatability of the response to selection over successive summers. High realized heritability for survival of selected lines, monitored separately or mixed together, confirmed the results obtained in the first generation. The relative survival performance of inbred lines over 3 successive generations of inbreeding tested in Ronce was highly repeatable. Survival during the second summer was generally high, except in Baie des Veys where high selected lines ("R") showed better survival than low selected lines ("S"), indicating a positive genetic correlation between survival in the first and second summers. Survival of selected lines was also tested under nursery conditions. In most cases, survival rates under nursery conditions were highly correlated with results from the field. No correlated response was observed between survival and growth in the field, indicating the absence of a genetic trade-off between these two traits. Alternatively, relationships between reproductive allocation and survival (i.e. trade-offs) might explain why a high genetic variance for survival appears to be maintained in the wild. Overall, our results strongly suggest that the establishment of a family-based breeding program to improve spat survival would be successful.