Based on indoor simulated rainfall test, this study compared the influence of intermittent rainfall and continuous rainfall on rill erosion on loess slope, quantified and compared the experimental errors of intermittent rainfall relative to continuous rainfall conditions, and evaluated the suitability of this experimental method in slope erosion simulation. The results showed that: (1) At low rainfall intensity (≤60 mm/h), for the intermittent rainfall test, the rill sediment production on the slope was significantly reduced with an error of -40.8%, and the error was between -33% ~ -45% along with the decreasing of ditch width, ditch depth and cleavage degree. The error of runoff production was small (0.4%). (2) When the rainfall intensity was high at 90 mm/h and 120 mm/h, the errors of various indexes of sediment and runoff production and rill morphology were all small, ranging from 3% to 12%, which had relatively little impact on the test results. (3) When the intermittent rainfall test method was used, it was suggested to adopt the high rainfall intensity test conditions (≥90 mm/h) to accurately simulate the rill erosion process of natural rainfall, and the rainfall duration of each stage of intermittent rainfall test should be controlled in case of high rain intensity, so as to avoid the continuous and rapid growth of accumulated sediment yield which leaded to the sharp development of rill erosion and further increase of error. Because of the difference of soil properties, the applicability of the test results in other areas needed further study. The results could provide technical support for the study of rill erosion process and mechanism.