In animal husbandry, there is a growing demand for animals that need less labor and are capable to handle diseases or other environmental challenges. For various markets, homogeneity of animal products is desired and uniform animals would ease management, e.g., less fluctuation in production in spite of challenges. Furthermore, due to globalization of breeding programs animals need to be capable to perform in a wide range of environments. Here we show 2 statistical approaches that can be used to breed for resilience and uniformity: a reaction norm model to breed for resilience and a double hierarchical generalized linear model (DHGLM) to breed for uniformity. Both were applied to reproduction traits such as total number of born piglets and number of piglets born alive in sow lines. For the reaction norm model, we first developed a challenge load indicator to estimate the level of challenge, based on drops in production. Subsequently, we used this challenge load indicator as a covariate in the reaction norm analysis. We found genetic correlations of 0.5–0.85 between healthy and diseased periods indicating substantial reranking of animals, or in other words genetic variation in resilience. We applied a DHGLM to total number born and found substantial genetic variation in residual variance of litter size with a genetic coefficient of variation at variance level of 0.17. Using deregressed variance EBV, we found a few highly significant genomic regions affecting the variance of litter size. These genomic regions could be utilized in genomic selection. Both statistical approaches can yield breeding values that could be used to select for increased resilience and uniformity of animal production. Due to its low heritability, accuracies of breeding values for resilience and uniformity are low, though substantial genetic variation is present. Accuracy of breeding values for these traits can be enhanced by genomic selection.