Due to shortage of facilities and tools, continuous monitoring of sediment load is not possible in most rivers and therefore the rating curve is usually used to estimate the sediment discharge. It has been proven that adding some other time related variables are necessary to improve the precision and accuracy of the rating curves. In this study, using a 7-parameter equation proposed by Cohen et al., multivariate sediment models for five hydrometric stations of Qazzaqli, Pounel, Keshvar, Gelinak and Bande Shah Abbasi, have been developed based on the recorded data during 1968– 2002. Then, these models have been evaluated using precision and accuracy indices. Furthermore, the effects of different watershed physiographic factors on the precision and accuracy of the rating curves has been investigated. Results showed that the average values of precision and accuracy indices across five stations were 1.958 and 3.488, respectively. This indicates the satisfactory performance of multivariate models in comparison with original single variable sediment rating curves. Also the mean annual flow discharge had the highest correlation with accuracy and precision indices which can be related to high frequency of its recorded data.