The C4 grass Echinochloa polystachya, which forms dense and extensive monotypic stands on the Varzea floodplains of the Amazon region, provides the most productive natural higher plant communities known. The seasonal cycle of growth of this plant is closely linked to the annual rise and fall of water level over the floodplain surface. Diurnal cycles of leaf photosynthesis and transpiration were measured at monthly intervals, in parallel with measurements of leaf area index, canopy light interception and biomass. By artificial manipulation of the light flux incident on leaves in the field light-response curves of photosynthesis at the top and near to the base of the canopy were generated. Fitted light-response curves of CO2 uptake were combined with information of leaf area index, incident light and light penetration of the canopy to estimate canopy rates of photosynthesis. Throughout the period in which the floodplains were submerged photosynthetic rates of CO2 uptake (A) for the emergent leaves were high with a mean of c. 30 μmol m-2 s-1 at mid-day and occasional values of 40 μmol m-2 s-1. During the brief dry phase, when the floodplain surface is uncovered, there was a significant depression of A, with mid-day mean values of c. 17 μmol m-2 s-1. This corresponded with a c. 50% decrease in stomatal conductance, and a c. 35% depression in the ratio of the leaf inter-cellular to external CO2 concentration (c i/c a). During the dry phase, a midday depression of rates of CO2 assimilation was observed. The lowest leaf area index (F) was c. 2 in November-December, when the flood plain was dry, and again in May, when the rising floodwaters were submerging leaves faster than they were replaced. The maximum F of c. 5 was in August when the floodwaters were receding rapidly. Canopy light interception efficiency varied from 0.90 to 0.98. Calculated rates of canopy photosynthesis exceeded 18 mol C m-2 mo-1 throughout the period of flooding, with a peak of 37 mol C m-2 mo-1 in August, but declined to 13 mol C m-2 mo-1 in November during the dry phase. Estimated uptake of carbon by the canopy from the atmosphere, over 12 months, was 3.57 kg C m-2. This was insufficient to account for the 3.99 kg C m-2 of net primary production, measured simultaneously by destructive harvesting. It is postulated that this discrepancy might be accounted for by internal diffusion of CO2 from the CO2-rich waters and sediments via the roots and stems to the sites of assimilation in the leaves.