In many Asian countries the future will demand a higher rate of rice production increase than that obtained in the 1970's. However, low rice prices discourage production and lead to rice shortages. There is a strong interest in identifying rice genotypes which effectively utilize resources on the land, and in developing low-cost production technology for tropical Asia. A method proposed earlier to evaluate N utilization efficiency of 24 rice genotypes, and initially tested on a soil with high N fertility, was tested on a soil with low N fertility in order to evaluate the performance of the genotypes under contrasting conditions. In six consecutive seasons on an infertile Vertic Tropaquept soil in Maligaya, Nueva Ecija, The Philippines, field performance was ranked on the basis of 12 parameters considered in the initial study. Significant differences among genotypes were observed in 68 out of 72 instances during the six seasons. Five parameters were selected for ranking in unfertilized plots and seven for ranking in plots receiving N fertilizer. Principal-component analysis showed that yield was the most significant single parameter in most instances. A two-parameter combination involving yield and ratio of panicle weight to either total N uptake or soil N uptake, however, correlated much better with rankings based on five or seven parameters. In contrast to the IRRI data on a fertile Andaqueptic Haplaquoll soil, significant differences among genotype means for soil N uptake in fertilized plots and total N uptake in unfertilized plots were found in all six seasons at Maligaya. Genotypes 11 and 18 ranked consistently high with all combinations of parameters, whether fertilized or unfertilized. For example, genotype 11 ranked first in all cases at IRRI, and either first or second in all cases at Maligaya. Others such as genotypes 5 and 12 ranked consistently low. The results indicate consistency in rice-genotype performance on soils deficient in N supply as well as those adequately supplied, in both wet and dry seasons, thus extending the applicability of the methodology.