Genetic improvement programs can increase the productivity of cultured aquatic species. Their rigorous design requires knowledge about parameters such as herit-ability and correlations among the traits under selection. These parameters enable the estimation of the breeding value of the selection candidates in the population, as well as the prediction of the potential response to selection. In Tilapia the focus of selection programs has been almost exclusively restricted to growth rate. Several estimates of heritability, in particular for live weight and growth rate, can be found in the literature (e.g. Kronert et al., 1989; Oldorf et al., 1989; Gall and Bakar, 2002; Bolivar and Newkirk, 2002). In a strict sense, such genetic parameters are only applicable to the population and the environment where they were obtained. Further¬more, individual estimates are subject to sampling problems and the parameters can change over time, particularly in relatively small populations undergoing selection. Hence, the desirability of having parameter estimates that are directly relevant to the population one is working with. In this paper we present an estimate of heritability for live weight (at approx. seven months of age) in two grow-out environments (cage and pond) for the GIFT (Genetically Improved Farmed Tilapia) strain (Eknath et al., 1993; Bentsen et al., 1998; Eknath and Acosta, 1998). We also estimate the response to selection in harvest weight by three different methods. The effects of environment (cage or pond) and sex on live weight are also discussed.