In this study, validity and limitation of the estimation of roughness coefficient using the measured field data are investigated and the errors of the calculated roughness coefficient are analyzed. The assumption of uniform flow led to much difference of the computed results in low flow, and this is due to change of the cross-section informations such as flow area and hydraulic radius rather than the difference of velocity head. From the comparison between the estimations of average roughness coefficient in the reach which is relatively long, the calculation using the modified Newton-Raphson method is very efficient and accurate. In the measured roughness coefficient, the errors of measured flow and stage are included and the lower flow is, the larger the magnitude of error of measured roughness coefficient is. But the error of depth and velocity associated with uncertainty of roughness coefficient is less than about 5% in the both of low and high flow, and it shows the validity of measured roughness coefficient.