The objective of this study was to investigate the relationship between soil detachment rate and sediment concentration of ridge contour during maize growing stages. This paper took sloping farmland of purple soil in Hilly Area of Central Sichuan basin as the research object. The filed artificial rainfall experiment was carried out for determination of characteristics of soil detachment rate and sediment concentration during maize growing stages

and to find the relationship between soil detachment rate and sediment concentration. The results showed that: (1) The soil detachment rate increased with the increasing rainfall intensity during maize growing stages

and the soil detachment rate under 2.0 mm/min rainfall intensity were 1.87 and 4.59 times greater than that under 1.0 and 1.5 mm/min rainfall intensity

respectively. During maize growing stages in the field

the soil detachment rate showed a falling-and-rising trend

and was significantly higher in the seedling stage than that in other growing stages under the same rainfall intensity. (2) Within maize growing stages

the sediment yield increased firstly and then decreased under different rainfall intensities. The largest sediment yield was (306.6 g) in the seedling stage

and the least (64.6 g) was in tasseling stage. The sediment yield increased with the increasing rainfall intensity in growing stages except for tasseling stage

while the difference was significant among rainfall intensities in seedling stage and jointing stage. The sediment concentration in runoff showed a fluctuating trend during the maize growing stages in the rainfall process

and the fluctuation range increased with the increasing rainfall intensity. (3) A power function relation was listed between soil detachment rate and sediment concentration in runoff during maize growing stages

and the correlation was significant. The results provide references for revealing the mechanism of soil erosion in sloping farmland of purple soil and the construction of regional water erosion prediction model.