In the population dynamics of bi-phasic marine invertebrates, the fitness of one stage in the life of a cohort affects that of the following stage. This effect makes the prediction of the fate of a cohort challenging. We conducted a 22 yr monthly survey of a population of cockles Cerastoderma edule focussed on the juvenile stage. Two types of years were distinguished in terms of cohort longevity: a low-mortality group (‘L’) and a high-mortality group (‘H’). The cockle shell lengths of the 0+ cohorts was higher in the L group, in August and September. A cohort could hereby confidently be assigned to the L or H group as soon as August. Maximum cockle abundance in the cohorts, parasite load, or date of recruitment did not discriminate the groups, while an early date for the peak of the 0+ cohort abundance could be related to the L group. The maximum air temperature and the chl a concentration in July were higher in the H group, whereas the minimum salinity during the month of September was slightly lower in the H group of years. Therefore, a juvenile shell length under a given threshold was identified as an early alert for a short lifespan of the cockle cohort. This long-term analysis contributes to the deeper understanding of the population dynamics of bi-phasic invertebrates. The success of juveniles is not solely related to the intensity of recruitment or mortality by predation, but could also be linked to their fitness, their growth rates appearing as a proxy.< Réduire