This study was designed to test for the presence of genotype × environment interaction (G × E) in Nile tilapia. We used 2 lines of Nile tilapia that had been divergently selected for two generations in two earthen pond environments. One line was selected in earthen ponds which were fertilized daily with chicken manure and which received no supplementary feeding (LL). The other line was selected in ponds receiving 25% protein pellets (HL). The 2 lines were tested in five test environments: 40% protein pellets feed (P200), 25% protein pellets (P100), 16% protein pellets (P50), 50 kg/ha chicken manure (M100) and 25 kg/ha chicken manure (M50). Nitrogen input was similar in P50 and M50, and in P100 and M100 treatments respectively. Survival to harvest ranged from 70 to 74.7% in the HL and from 62.3 to 76.3% in the LL. Treatment, pond within treatment, sex and age effects, initial weight, and the interaction between sire and treatment significantly affected body weight at harvest in both lines. Heritability estimates between treatments within each line were highly variable, with high stand errors. Common environment estimates varied between 0.18 and 0.44. Genetic correlations between treatments ranged from 0.76 to 0.99 in the HL and from 0.57 to 0.99 in the LL. We found low genetic correlations (0.39 to 0.56) between the environment of selection and the test environments. Together these results indicate the existence of G × E in Nile tilapia.