Time series analysis is a very effective method to analyze the dynamic characteristics of soil moisture at long-term scale. In this study, we have used the time series to analyze the relationship between precipitation and soil moisture on fixed dune at different soil depths (from 0 to 120 cm) during the growing season (from May to September) of 2006-2010 in Korqin Sandy Land, northern China. The results indicate that: (1) The precipitation is a relatively independent time series and has no obvious autocorrelation. Precipitation in an earlier stage has no obvious effect on the later stage in the growing season. (2) Soil moisture in different soil layers has higher autocorrelation; and the autocorrelation of soil moisture in each soil layer is significantly weakened with the increase in time lag interval. (3) The correlation coefficient between soil moisture and precipitation in each soil layer is higher at the time lag interval of k = 0; with the increase in soil depth, the correlation is gradually weakened. (4) The maximum correlation coefficients of soil moisture series and precipitation series in different soil depths were obtained at the time lag interval of k = 0.