Both phenotypic and genetic quality can influence the survival of individuals through time, although their relative influences are rarely addressed simultaneously. Here we used capture-mark-recapture modelling to assess the influence of both multilocus heterozygosity (MLH) and body size on apparent adult survival in a tropical bird species, the Zenaida dove, Zenaida aurita, using a sample of 391 individuals genotyped at 11 microsatellites, while controlling for the effects of sex. No effect of body size on either adult survival or capture rate was found. In the best model, survival was a logit linear function of MLH, whereas detection probability was a sex-dependent logit linear function of the logarithm of field effort, increasing with time and affected by a random individual effect. Using a Bayesian approach, we found that MLH explained 1.14% of the total deviance, as expected from theory and previous studies of heterozygosity-fitness correlations, with no evidence for local effects. However, results from capture-mark-recapture modelling indicated that expected longevity varied from 4.8 years in the least heterozygous individuals (MLH = 0.37) to 10.6 years in the most heterozygous ones (MLH = 1), thus suggesting that MLH had potentially a substantial effect on survival. We discuss our results in relation to current hypotheses about the origin of heterozygosity-fitness correlations.