Generating breeding programmes that effectively improve farmed fish performance across multiple environments and make fish more uniform within production environments would aid farmers to produce food under diverse environments. We review genotype-by-environment interaction leading to re-ranking of genotypes across environments, that is non-unity genetic correlation between traits measured in different environments, and micro-environmental sensitivity leading to a change in environmental variance of a trait. A quantitative review across 38 species showed that (i) genotype-by-environment interaction studies are lacking for many economically important traits. (ii) Re-ranking is moderate for growth (average genetic correlation = 0.72) and survival (average genetic correlation = 0.54). Significant re-ranking is of concern because selection in a nucleus leads to lower genetic responses in commercial environments compared to a case when re-ranking does not exist. (iii) Re-ranking is weak for age-at-sexual-maturity and fish appearance (average genetic correlation = 0.86), implying that genetic improvement in one environment is expected to be effective in the other environments. Future research should provide guidelines for how to account for genotype-by-environment interaction when collecting data, estimating breeding values and optimising the structure of the breeding programme. (iv) Coefficient of genetic variation for sensitivity against unknown micro-environmental factors within a single environment for body weight is high. Hence, genetic improvement towards less sensitive fish, resulting in more uniform production, is possible, but a large number of relatives with phenotypes is needed for obtaining moderate accuracy of selection. This review elucidates needs for further research on genotype-by-environment interaction and micro-environmental sensitivity in economically important traits and species.