A study on diurnal variations of CO₂, CH₄ and N₂O fluxes in the late growth stage of wheat under conventional tillage and no-tillage were conducted at the sixth year of a long term located experiment. The Gas Chromatography (GC), ascertained the optimal observation durations for CO₂, N₂O and CH₄ fluxes were investigated for 48 hours using the Static-Chamber method. The results showed that the CO₂, CH₄ and N₂O fluxes had significant diurnal variations under both conventional tillage and no-tillage treatments, which it might be a sink of CH₄ but a source of N₂O and CO₂. The diurnal flux of CO₂ emission peaked in PC 752.58 mg·m^-2·h^-1 at 12 p.m., and the lowest value was AC 186.42 mg·m^-2·h^-1 at 2 a.m.; while the N₂O peak fluxes of AC, PC and PZ presented were performed at 2 p.m., 4 p.m., 4 p.m., respectively, and the peak fluxes values of AC, PC and PZ was 45.85 μg·m^-2·h^-1, 41.89 μg·m^-2·h^-1, 34.38 μg·m^-2·h^-1, respectively, however, N₂O flux of AC decreased rapidly after 6 p.m. and presented the lowest value at 0 a.m. with 16.69 μg.m^-2.h^-1, but PC and PZ was at 4 a.m., the similar trend was noted in CH₄, but there were two flux peaks in a day, which observed at 2 p.m. and 0 a.m., and the lowest flux observed at 6 a.m.. The daily mean emission fluxes of CO₂ was PC > AC > PZ, CH₄ was AC > PC > PZ, but N₂O was PC > AC > PZ. There were highly significant (P<0.01) or significant (P<0.05) positive correlations between gas fluxes and land surface temperature or soil temperature of 5cm depth under conventional tillage and no-tillage systems, and the temperature might be a crucial factor in diurnal variation. The percentage of CO₂, N₂O and CH₄ fluxes in daily showed that CO₂, N₂O and CH₄ fluxes in daytime was 74%, 66% and 61%, respectively, further, the ratio between daytime and night presented that CO₂ was 3 to 1, N₂O was 3 to 2 and CH₄ was 2 to 1, there are significant fluxes difference between each gases. In order to make sure the optimal observation duration for this periods and areas, we calculated the correction coefficient and regression analysis between the observation duration and the daily total fluxes of three gases, which showed that the correction coefficient of CO₂ flux on the verge of 1 from 8 a.m. to 10 a.m., and the related coefficient was 0.8175 (P<0.01), it's showed that this time period significantly represented daily mean flux of CO₂ or it was the optimal observation duration for CO₂ collection in this period and area. The optimal observation duration for N₂O was 10 a.m. to 12 a.m. and the related coefficient was 0.7642 (P<0.05), but two values for CH₄ were 8 a.m. to 10 a.m. and 0.8396 (P<0.01), respectively, which correspondingly presented the representative emission flux of a day and was the optimal observation duration of the late growth stage of wheat for greenhouse gases in this area.