The main aims of this study were to predict the increase in the accuracy of selection and the response to selection achieved by including feeding pattern traits, such as number and size of meals, as selection criteria in indices to improve growth rate, lean content of the carcass and food conversion ratio. Genetic and phenotypic parameters obtained for pigs given food ad libitum using single space electronic feeders were used to construct the indices. The predicted genetic gain in the index increased for indices with more feeding pattern traits. The inclusion of part test records for feeding patterns resulted in similar predicted correlated responses and predicted accuracy of selection to those using whole test records of daily food intake. The inclusion of feeding patterns as selection criteria resulted in indices that were less robust to inaccurate parameter estimates. It was concluded that feeding pattern traits could be used to improve the predicted accuracy of selection for the efficiency of lean growth but the most effective and robust index would include only daily gain, backfat depth, daily food intake and mean number of visits to the feeders per day. These traits may also be useful in reducing the length of the test period necessary for accurate measures of food intake, so increasing the potential intensity of selection using a given number of feeders.