1. Heterogeneity in individual quality (i.e. individuals having different performance levels that are consistent throughout life) can drive the demography of iteroparous species, but quality in the context of environmental variability has rarely been evaluated. 2. We investigated the demographic responses of a long-lived seabird, the Adélie penguin (Pygoscelis adeliae), to contrasting environmental conditions as a function of reproductive success, breeding quality (BQ) and experience. A continuous index of BQ (BQI) was developed to reflect an individual's ability, relative to others, to produce viable offspring. 3. First, we assessed the relative importance of costs of reproduction vs. heterogeneity in quality by comparing survival and reproductive probabilities among deferred, successful and unsuccessful breeders under 'demanding' conditions using multistate capture-mark-recapture modelling. Then, we quantified the influence of BQI on adult survival among experienced breeders vs. the whole study population under both 'normal' and 'demanding' conditions. 4. Higher survival rates were exhibited by successful (74-76%) compared to unsuccessful breeders (64%); the former also more frequently reproduced successfully at year t + 1. 5. From 1997 to 2006, adult survival ranged from 64-79%, with BQI accounting for 91% of variability in the entire study population, but only 17% in experienced breeders. The weakened relationship between BQI and survival in experienced breeders supports the theory that selection during the first reproductive event accounts for a more homogeneous pool of experienced breeders. 6. No significant effect of environmental covariates on survival was evident, suggesting that what appeared to be demanding conditions were within the range that could be buffered by this species. 7. For the first time in seabirds, a quadratic relationship between adult survival and BQI showed that adult survival is shaped by both heterogeneity in quality and reproductive costs. Our study confirms that population demographic patterns are affected by factors at the individuals' level (e.g., individual quality) that are obscured at population-scale levels.