Growth models are used commercially to determine feeding strategies that improve profitability on pig farms. A simulation study was conducted to determine the impact of genotype (fat, normal and lean), population size (N = 1, 5, 25, 125 and 625), variance of certain genotype parameters (coefficient of variation CV = 0, 5, 10, 20%) and covariance of genotype parameters (maximum protein deposition, minimum body lipid to protein ratio, digestible energy intake) on the feeding strategy which maximises gross margin. The maximum gross margin was determined using a genetic algorithm and each combination was simulated 10 times. Profitability and lysine to energy ratio in the diet were higher for the lean genotype than for the normal and fat genotypes. For a given population size, increasing the variation between pigs increases the variability in gross margin and, for a given level of variation between pigs, the variability in gross margin decreases as the population size increases. Taking into account the covariance between pig parameters results in higher lysine to energy ratio in the diet. Feeding the diet which maximises gross margin for a single pig to a population of pigs results in lower gross margin; this is true for all the genotypes. These results illustrate the economic advantage of using a stochastic growth model with covariance when determining optimal diets for pigs.