The permanent measurement of discharge even in normal conditions is difficult and very expensive and

the related problem is increased when floods take place. Towards solving the problem, it is tried to develop

discharge-stage relationship through reading the stage and measuring the associated discharge in particular

conditions. In this study, the stage-discharge relationship was developed and then analyzed for different

time scales of rising and falling limbs of hydrographs, daily, monthly and seasonal by using bivariate

regression and classic discharge rating models. The study was conducted in Kojour Forest Educational

Watershed with an area of some 50000 ha in Mazandaran Province, Iran. The results showed that the developed

relationships, for each time scale, are specific, so that they cannot be applied for the other scales with

reliable accuracy. The linear and power regression models with the minimum correlation coefficient of

58%, the maximum respective calibration and verification relative errors of 70 and 75%, lower RMSE and

the minimum efficiency coefficient of 35% were also found superior to and more accurate than classics

models for all study time scales.