The effect of soil and water conservation measures (SWCMs) is usually dependent on time. Thus the trend in reducing runoff and sediment over time is a very important theoretical problem for evaluating the effectiveness of SWCMs. Moreover, there is still a lack of comprehensive assessment of water erosion dynamics following implementing SWCMs despite their ecological significance. Therefore, the long-term impact of SWCMs on runoff and sediment and their relationships was assessed for an orchard on sloping red soil in southern China. Overland flow and erosion sediment were continuously observed for 15 years on citrus experimental plots under one of four treatments: grass strips, strip intercropping, level terrace and clean-tillage. By means of Mann-Kendall trend tests and double cumulative curves, the time series of runoff and sediment under the different treatments were analyzed. Furthermore, we linked the effect of soil conservation and the relationship between runoff and sediment variation to determine the mechanism of conservation measures on sediment reduction. The results showed that the first 4 years was the key period to prevent soil erosion for this orchard, and then the intensity of soil erosion decreased below 500 t·km-2·a-1. Considering economic costs and ecological effect, grass strips were the best protective measure for this test situation. The fitted curves of the effect of SWCMs on sediment reduction over time showed an 'L' form, but on runoff there was an approximately horizontal line. The SWCMs did not change the rainfall-runoff relationship, but did change the runoff-sediment erosion relationship. The erosion reduction mechanism of SWCMs in the early phase was a joint function of reducing runoff and changing the runoff-sediment relationship, and in the post-stable phase it worked mainly by reducing runoff. The results provide the basis for rational allocation of SWCMs considering location and time.