A large number of rainfed lowland rice genotypes were grown under two fertiliser conditions to identify whether a genotype’s ability to extract more nutrients or to use absorbed nutrients more efficiently to produce grain yield was more important in determining genotypic variation in grain yield. From the yield responses of lines to fertiliser application, 16 contrasting lines were selected, and dry matter and nutrient (N, P and K) contents were determined for grain and straw separately for crops grown under non-fertilised and fertilised (60–13–16 N–P–K kg ha −1) conditions at three locations in Laos. There were significant effects of both genotype and genotype-by-fertiliser interaction for grain yield, which were closely associated with total N and P content at maturity. There was, however, also significant genotypic variation in nutrient-use efficiency (grain yield per unit nutrient absorbed), and this also contributed to the genotypic variation for grain yield. There was a large effect of genotype and a smaller effect of genotype-by-location interaction for nitrogen-use efficiency for grain yield. Nitrogen-use efficiency was negatively correlated with grain N concentration at two of the three locations. Both N- and P-use efficiency were consistent across fertiliser levels, and hence are likely to be used as selection criteria. The work also indicates that genotypes with high harvest index (HI) are likely to perform well in different fertility conditions.