Abstract: Abstract: Gully erosion is one of main forms of soil erosion threatening land resources. Conservation practices that can reduce rill and inter-rill erosion would be effective on control gully erosion but there is a lack of quantitative assessments. In order to access the effect of terrace on gully erosion, 32 small watersheds were selected randomly on the hilly and gully loess plateau in northern Shaanxi province, where cropland was returned to woodland or grassland in 2001 and terrace was the major conservation practices. Based on two QuickBird images with a 9-year interval and digital elevation model (DEM) with pixel size of 5 m, valley shoulder line and distribution of terrace and land-uses in the small watersheds was extracted by visual interpretation and with GIS technology. Average vegetation cover, local slope upslope drainage area and changes of the gully area from 2004 to 2013 were calculated to access the factors influencing gully growth. Gully area growth with cropland on terrace, woodland on narrow terrace, or grassland on narrow terrace on the upslope drainage area and gully area growth without terrace on the upslope drainage area were compared with statistical analysis method. The results showed that the proportion of terrace area on the upslope drainage area in the 32 small watersheds increased from 39.43% to 59.39% during the period from 2004 to 2013 with woodland on narrow terrace increased by 11.76% on average, cropland on terrace increased by 9.91% and grassland on narrow terrace decreased slightly. The gully area in 32 investigated small watersheds in 2013 increased by 0 to 6% with the average of 4.18%, as compared to 2004. Among the three factors related to gully growth rate investigated in this study, the terrace proportion in upslope drainage area was the most important. Gully area growth rate correlated significantly with terrace proportion in upslope drainage area and average vegetation coverage at the level of 0.01, with the Pearson's correlation coefficients were -0.801 and 0.671, respectively; while the geomorphologic factors (AS2, where S is local slope gradient and A is drainage area), which was considered the predictor of gully initiation, were not related significantly to gully area growth rate owing to high vegetation coverage and high terrace proportion on the upslope drainage area, which reduced the kinetic energy of concentrated overland flow. 32 small watersheds were classified into 5 groups according to terrace proportion and vegetation types. Gully area growth rate in 5 groups of small watersheds increased in the following order: woodland on narrow terrace＞grassland on narrow terrace＞cropland on terrace＞multiple vegetation on terrace＞multiple vegetation without terrace. It indicates that the combination of engineering measures such as terrace and conservation vegetation achieves better effect than the single one. The results of this study can provide valuble information for evaluating the full effect of terrace on controlling soil erosion and making plan for gully control.