A recent breakthrough in generating fertile progeny from Oryza sativa X 0. glaberrima crosses gives rice breeders access to a broader range of germplasm. Interspecific crosses might provide new solutions to the low productivity of upland rice systems prone to weed competition. Two field and one pot experiments conducted during 1995 and 1996 served to characterize growth and yield potential of CG14 (0. glaberrima), WAB56-104 (0. sativa) and their progeny. During the 1995 wet season and the 1996 dry and wet seasons, the lines were seeded in a well-drained upland field in Ivory Coast with supplemental sprinkler irrigation. A randomized complete-block design with three replications was used, with cultivar and nitrogen levels as sub-plots. Specific leaf area (SLA), leaf area index (LAI), leaf chlorophyll content (SPAD method) and tiller number were measured at 2-week intervals until flowering. Grain yield and yield components were measured at maturity. In all environments, CG14 produced two to three times the LAI and tiller numbers as WAB56-104. This was associated with a high SLA and low leaf chlorophyll content. Grain yields of CG14 did not respond to N inputs, although the sink potential did. The difference was mainly caused by grain shattering. The progenies had intermediate LAI, SLA and leaf chlorophyll content, but their grain yields, tiller numbers and resistance to lodging and grain shattering were similar to WAB56-104. Across lines, LAI and SLA were significantly correlated. A paddy field experiment confirmed the relationship between LAI and SLA for a wider range of rice cultivars and interspecific progenies. A pot experiment demonstrated that leaf net CO, assimilation rates (A...) followed a common linear function of area] leaf chlorophyll content across cultivars. The main common cause of differential LAI and A.. appeared to have been genotypic patterns of SLA, which might be an important determinant of growth vigor and competitiveness with weeds. The possibil