At ecosystem level, water use efficiency (WUE, defined as the net carbon uptake per amount of water lost from the ecosystem) reflects the coupling of carbon and water cycles. Continuous observation of carbon and water fluxes was made over Horqin Meadow with eddy covariance system in growing season of 2012. This paper calculated the water use efficiency, WUE_GPP, defined as the ratio of gross primary productivity, GPP, to evapotranspiration, ET. Diurnal and seasonal variation of WUE_GPP and its response to environmental and physiological factors were analyzed. The results showed that the diurnal variation of WUE_GPP presented a trend of decreasing-stable-increasing, and the appearance of maximum was one or two hours later after sunrise. WUE_GPP on cloudy day was greater than that on sunny day, and the WUE_GPP at the mid of the growing season was greater than that in the beginning and at the end of the growing season. Seasonal variation of GPP and ET both presented a trend of higher in summer and lower in spring and autumn. Their maximum values appeared at the end of June and in mid-July, and were 1.49 mg m^-2 s^-1 and 0.16 g m^-2 s¹, respectively. Because of harvest, GPP and ET decreased rapidly around September 1st. The seasonal averaged GPP and ET were 0.57 mg m^-2 s^-1 and 0.08 g m^-2 s¹ respectively. Seasonal variation of WUE_GPP also presented a trend of higher in summer and lower in spring and autumn, and the maximum value (13.62 mg/g) appeared at the end of June. The seasonal averaged WUE_GPP was 5.97 mg/g, which was higher than that of forest ecosystem and lower than that of desert grassland ecosystem. GPP presented a quadratic relationship with vapor pressure deficit, air temperature and leaf area index, respectively, and a logarithmic relationship with canopy conductance. ET presented a quadratic relationship with air temperature, while its correlation with vapor pressure deficit, leaf area index and canopy conductance was non-significant. WUE_GPP presented a quadratic relationship with vapor pressure deficit, air temperature and leaf area index, respectively, and a logarithmic relationship with canopy conductance. Vapor pressure deficit equal to 2.0 kPa, leaf area index equal to 4.2 and canopy conductance equal to 0.002 m/s were the critical points that limit the increase of WUE_GPP. When vapor pressure deficit was less than 2.0 kPa, WUE_GPP enlarged with the increase of vapor pressure deficit, and when it exceeded 2.0 kPa, the situation was opposite. With the increase of leaf area index, WUE_GPP enlarged gradually under low leaf area index, but when it exceeded 4.2, WUE_GPP did not increase and tended towards stability with the continual increase of leaf area index. With the increase of canopy conductance, WUE_GPP enlarged rapidly at begin, but slowly with the continual increase of canopy conductance when it exceeded the critical point. Water use efficiency expressed by other productive indexes was also estimated, including WUE_NEP (defined as the ratio of net ecosystem production, NEP, to ET) and WUE_NPP (defined as the ratio of net primary production, NPP, to ET). The seasonal variation of WUE_NEP and WUE_NPP presented a similar trend with WUE_GPP, and their averages were 3.47 and 5.47 mg/g respectively. Comparison of WUE in different ecosystems found that leaf area index was the dominant factor of WUE magnitude because that leaf area index was a comprehensive reflection of air temperature and precipitation, while air temperature or precipitation alone can not determine the WUE of an ecosystem.