Residual feed intake (RFI) is defined as the difference between the observed average daily feed intake (ADFI) and the ADFI predicted from production and maintenance requirements. The objectives of this study were to evaluate RFI as a selection criterion to improve feed efficiency and its potential to reduce Nitrogen (N) and Phosphorus (P) excretion in four pig breeds. Data were collected between 2000 and 2009 in French central test stations for two dam breeds (French Landrace (LR) and Large White (LWD)), and two sire breeds (Large White (LWS) and Piétrain (PP)). Numbers of recorded pigs were 6407, 10694, 2342 and 2448 for the LR, LWD, LWS and PP breeds, respectively. All PP animals were genotyped for the halothane mutation. This data set was used to calculate RFI equations for each of the four breeds, and to estimate genetic parameters for RFI together with growth, carcass and meat quality traits, and N and P excretion during the test period (35-110 kg body weight). The RFI explained 20.5% in PP, 26.5% in LWS, 27.6% in LWD and 29.5% in LR of the phenotypic variability of ADFI. The PP breed differed from the others in this respect, probably due to a lower impact of the variation of body composition on ADFI. Heritability estimates of RFI ranged from 0.21 ± 0.03 (LWD) to 0.33 ± 0.06 (PP) depending on the breed. Heritabilities of N and P excretion traits ranged from 0.29 ± 0.06 to 0.40 ± 0.06. The RFI showed positive genetic correlations with feed conversion ratio (FCR) and excretion traits, these correlations being greater in the sire breeds (from 0.57 to 0.86) than in the dam breeds (from 0.38 to 0.53). Compared to FCR, RFI had weaker genetic correlations with carcass composition, growth rate, and excretion traits. Estimates of genetic correlations between FCR and excretion traits were very close to one for all breeds. Finally, excretion traits were, at the genetic level, correlated positively with ADFI, negatively with growth rate and carcass leanness, whereas the halothane n mutation in Piétrain was shown to reduce N and P excretion levels. To conclude, new selection indexes including RFI can be envisaged to efficiently disentangle the responses to selection on growth rate and body composition from those on feed efficiency, with favorable impacts on N and P excretions, particularly in sire pig breeds. However, the switch from FCR to RFI in selection indexes should not allow to resolve the genetic antagonism between feed efficiency and meat quality.