Early life conditions, especially in long-lived organisms, can have both immediate and long-lasting effects in vital traits generating demographic structure across cohorts. Multiple non-exclusive hypotheses have been proposed to explore this question. For instance, the silver spoon, the viability selection or the predictive adaptive response hypothesis, predict that long lasting effects resulting from harsh early conditions could be negative, positive or vary with current environmental conditions, respectively. We use an 18-year capture–mark–recapture dataset on adult Audouin’s gulls Ichthyaetus audouinii to test for these different hypotheses while accounting for age, breeding experience and large-scale dispersal. Audouin’s gull cohorts experiencing harsh conditions during early life (i.e. nestling period and first winter) are known to experience lower first year survival. Here, we show that early life conditions also explained a large proportion (54%) of adult survival variation among cohorts. However, adulthood cohorts experiencing poor early life conditions had higher adult survival, in accordance with the viability selection hypothesis. Our results also show that apparent inexperienced breeders showed lower survival than experienced ones. Moreover, adult survival decreased with age. These results could suggest an increased cost of reproduction for deferred breeders, individual quality differences or survival senescence in this population. Overall, our study highlights the importance of early development, age and breeding experience as potential factors generating heterogeneity of survival between cohorts. Understanding the mechanisms driving responses to early life conditions at different life stages is fundamental to understanding the long-term dynamics of wild populations.