Estimating suspended particulate matter (SPM) fluxes in rivers with low uncertainty requires continuous time series of both discharge (Q) and SPM concentration (Cs). Unfortunately sediment load measurements are often infrequent and power-rating curves are usually used to simulate Cs continuous data from Q recordings. At five stations of three large French streams equipped with hydroelectric works, continuous data of Q and infrequent data of Cs were recorded. Continuous monitoring of the turbidity for several years was also available for the five stations. As a good agreement between the turbidity and the Cs was verified, we used the turbidity as a proxy of Cs to get an insight of its time dynamics during floods. A poor correlation between Q and Cs was obtained. Pronounced Cs peaks were observed during the Q rising limbs, leading to hysteresis phenomena. These observations suggest that concentration peaks in these rivers primarily stem from the resuspension of fine material stored in upstream reaches. We assessed the performance of two analytical models derivated from the power regression method to better simulate continuous Cs time series.The first method introduces the lag time between the discharge peak and the Cs peak. The second method calculates Cs as a function of the discharge Q and of the discharge change rate dQ/dt. In addition to the original power rating, these two methods were calibrated and tested to reproduce Cs time series at the event-scale and to estimate SPM fluxes at a longer timescale. Our final perspective is to develop a more physical approach of Cs as a function of the difference between aggradation and erosion of a sediment stock located upstream, using the bottom shear stress as a controlling parameter.