Soil respiration (Rs) is an important component of an ecosystem's carbon cycle and the main pathway for carbon release from the ecosystem into the atmosphere. The transfer of carbon between the ecosystem and atmosphere is of interest in the study of greenhouse gas emission. In order to understand characteristics of the Rs dynamics and the factors controlling it, this study was conducted in five typical vegetation types in Tianlaochi catchment area of the Heihe River, located in Qilian Mountains, Northwestern China. Rs, soil temperature, and soil moisture were measured by an automated soil CO₂ flux system (LI-8100) from May to September in 2013, and weather data for the corresponding study period were obtained from the weather station located in the experimental plot. The results showed that Rs in the five vegetation types displayed clear diurnal dynamic pattern. The Rs rate differed in different vegetation types, from the highest in subalpine steppe (2.01 to 12.10 μmol m^-2 s^-1), followed by Stipa purpurea steppe (1.40 to 8.45 μmol m^-2 s^-1), subalpine shrub (0.85 to 9.24 μmol m^-2 s^-1), Sabina przewalskii forest (1.09 to 4.14 μmol m^-2 s^-1), to the lowest in Picea crassifolia forest (1.25 to 3.19 μmol m^-2 s^-1). The daily change of Rs followed the fluctuations in surface air temperature, and there was a hysteresis between Rs and soil temperature. Rs rate had obvious seasonal variation, increasing from May and reaching the maximum in July, and then decreasing until the final measurements in September. The statistical analysis indicates that Rs rate was significantly positively correlated with soil temperature, surface air temperature, and air temperature, which relationship can be expressed with an exponential function. In relation to the temperature, Rs rate was more correlated to surface air temperature and air temperature than to soil temperature. For example, surface air temperature explained 79%-95% of the variation of Rs rate during the observation period, and air temperature explained 50%-77%, whereas soil temperature in the top 10 cm layer explained only 23%-47%; Rs rate had a significantly negative correlation with soil moisture in five vegetation types with R² ranging from 0.55 to 0.93. Further analysis indicated the positive linear correlation between Rs rate and wind speed. Principal component analysis showed that soil temperature and moisture in the top 0-60 cm layer of the soil are the first-line impact factors for Rs rate in five vegetation types. The temperature and moisture in surface soil are the second main impact factor, and solar radiation is the third main factor to affect Rs rate. The results indicate that Rs was significantly different in different vegetation types, and changes in vegetation alter the pattern of the Rs rate. This study has important implications as it helps to understand the role that different vegetation types play in reduction of carbon emission. Such information will lay a foundation for assessing carbon source or carbon sequestration of different vegetation types in Qilian mountainous area.