A fully pedigreed population based on the sixth generation of GIFT (Genetically Improved Farmed Tilapia) was established in Malaysia in 2002. Progeny were generated in two spawning seasons, 2002 and 2003. A number of statistical models were fitted to the data collected throughout the study, either to estimate breeding values (EBVs), variance components, or response to selection. Parents used in the spawning season of 2003 were either selected as having high estimated breeding values for live weight (LW) at approx. 7 months of age, or as having EBVs as close as possible to the average. In this way a Selection and a Control line were created, respectively. Two production environments were used to grow-out the progeny. At approx. 7 months of age females' live weight was 84 per cent that of males, whereas live weight in cages was 83 per cent of that in ponds. The heritability estimated from the animal variance component was 0.31 (s.e. 0.069), whereas the maternal and common environment effect estimated from the dam variance component was 0.15 (s.e. 0.031). Response to selection was estimated by three methods. Expressed as a percentage of the overall least squares mean for LW in the population, the response was about 10 per cent. The results are discussed in relation to other work. It was concluded that there was still additive genetic variance in the GlFT population established in Malaysia, and that it was capable of further response to selection. The issue of genotype by environment interaction is briefly discussed, and it was concluded that there was no justification for the conduct of separate genetic improvement programs in cage and in pond environments.