Abstract. The seasonal change in CO₂ flux over an artificial grassland was analyzed from the ecological and meteorological point of view. This grassland contains C₃ and C₄ plants; the three dominant species belonging to the Gramineae; Festuca elatior (C₃) dominated in early spring, and Imperata cylindrica (C₄) and Andropogon virginicus (C₄) grew during early summer and became dominant in mid-summer. CO₂ flux was measured by the gradient method, and the routinely observed data for the surface-heat budget were used to analyze the CO₂ and H₂O exchange between the grassland and atmosphere. From August to October in 1993, CO₂ flux was reduced to around half under the same solar-radiation conditions, while H₂O flux decreased 20% during the same period. The monthly values of water use efficiency, i.e., ratio of CO₂ flux to H₂O flux decreased from 5.8 to 3.3 mg CO₂/g H₂O from August to October, the Bowen ratio increased from 0.20 to 0.30, and the ratio of the bulk latent heat transfer coefficient C_E to the sensible heat transfer coefficient C_H was maintained around 0.40-0.50. The increase in the Bowen ratio was explained by the decrease in air temperature from 22.3 °C in August to 16.6 °C in October without considering biological effects such as stomatal closure on the individual leaves. The nearly constant C_E/C_H ratios suggested that the contribution ratio of canopy resistance to aerodynamic resistance did not change markedly, although the meteorological conditions changed seasonally. The decrease in the water use efficiency, however, suggested that the photosynthetic rate decreased for individual leaves from August to October under the same radiation conditions. Diurnal variations of CO₂ exchange were simulated by the multi-layer canopy model taking into account the differences in the stomatal conductance and photosynthetic pathway between C₃ and C₄ plants. The results suggested that C₄ plants played a major role in the CO₂ exchange in August, the contribution of C₄ plants decreased in September, and daily variations of CO₂ exchange were mainly due to C₃ plants in October. The results also suggested that the decrease in the net canopy CO₂ exchange from August to October was induced partly by the decrease of net photosynthesis on the individual leaves in both C₄ and C₃ plants, which could be due to aging of the leaves.