The data used in these analyses were compiled from individual feeding records of 1832 pigs from 70 sire families using FIRE (food intake recording equipment) system from Hunday Electronics Ltd at the Cotswold Pig Development Company. Pigs were on test between 45 (s.d. 2·76) kg and 95 (s.d. 6·78) kg. Daily food intake (DFI kg), food intake per visit (FlV kg), number of visits per day (NV), duration of each visit (TV min), time in the feeder per day (TD min), feeding rate (FR kg/min) and number of non-feeding visits per day (NFV), were measured as means of test and DFI was also recorded as means of bi-weekly periods of test. Performance test traits ofbackfat depth off test (BF mm), food conversion ratio (ECR kg/kg) and average daily gain (ADG kg), over the test period, were also measured.
Parameters were estimated by restricted maximum likelihood with a multivariate individual animal model. DFI had a heritability of 0·21 ranging from 0·18 to 0·26 over the four test periods. Correlations between DFI in each test period were high (rg
= 0·75 to 0·99). DFI was highly correlated with performance test traits (0·61 to 0·78) but had low correlations with feeding pattern traits (0·0 to 0·24). The heritabilities of feeding pattern traits were low (0·06 to 0·11) with the exception of FIV (0·27) and NV (0·34) but correlations between feeding pattern traits were high. FIV, NV and TV were moderately correlated with ADG (rg
= 0·49, -0·29, 0·33 respectively), BF (rg
= 0·35, -0·15, 0·17 respectively) and ECR (rg
= -0·12, 0·31, -0·27 respectively). Feeding patterns may be changed substantially by selection and the genetic correlations with performance test traits indicate that feeding patterns traits can be usefully incorporated in selection criteria to improve somewhat the accuracy of selection.