Improved feed efficiency is one of the major objectives of animal breeding. Measurement of residual metabolizable energy intake (RMEI) has been proposed and studied as an alternative to the feed to gain ratio to evaluate feed efficiency in laying hens (Bordas and Merat, 1980), dairy cattle (Korver, 1988) and pigs (Mrode and Kennedy, 1993). The RMEI is the difference between actual energy intake and the estimated energy requirements for maintenance and production. Variation in RMEI between animals may result from differences in maintenance requirements or in the utilization of dietary energy for production. However, the variation in RMEI may also result from measurement errors or errors in the estimation of its components (Luiting, 1991). For instance, differences in animal’s activity might be an important source of variation in maintenance requirements (de Haer et al., 1993). Thus, Luiting (1991) demonstrated that hen lines selected for high RMEI produced more heat than those selected for low residual feed intake, resulting in part from their enhanced physical activity (between 30 to 50% higher). More recently Gabarrou and Géraert (1994) suggested that this increase in heat production could result from an increased basal metabolic rate, physical activity or diet induced thermogenesis. de Haer et al. (1993) found that RMEI was strongly correlated with feeding frequency and duration, indicating that RMEI increases when feeding activity increases. These authors indicated that feeding activity alone explains 44% of the variation in RMEI. According to these results it can be hypothesized that, between genotypes or between individuals, RMEI variation results from variation in animal behaviour, the latter being an important determinant of energy requirements for maintenance. In particular, differences in nervousness, aggressiveness and physical activity in pigs might be at the origin of differences in RMEI. In growing pigs from three genetic lines, namely Genex Meishan-derived dam line (GM), Synthetic Genex 3000 (SG) and purebred Large White (LW), RMEI was determined and related to behaviour by factorial Partial Least Squares analysis. Behaviour was measured by observations of daily behaviour in the pen, postures, feeding behaviour, aggressive behaviour after mixing and ease of handling at weighing. The RMEI obtained for the GM pigs was lower (P=0.003) than that of the two other lines studied (LWand SG). This difference may result from inaccuracy in the estimation of requirements. A small proportion of these differences were explained by differences in behaviour. Indeed, standing posture, time spent in the feeder, toy manipulation, aggressive and locomotor behaviour explained part of the variation in RMEI, presumably because an increase in these physical activities induces higher energy expenditure. Vocalisations during manipulations at weighing were the only behaviour related to stress that influenced RMEI.