In the present study, a monthly time series model was developed for pan evaporation (Epan) in Jorhat (Assam), northeast India. We assume that the Epan series can be decomposed into deterministic and stochastic components, and a turning point test was used to identify the presence of a trend component in the series. A trend-free Epan time series was further obtained by eliminating the trend component from the original Epan time series, and later used in identifying the periodic component using Fourier series analysis. Results show that the first three harmonics explained the total variance by 88,2 %. Before modelling the stochastic dependent component, the periodic component was removed from the time series. In the stochastic dependent component modelling, the moving average (MA) model of order 2 was found suitable on the basis of minimum value of the Bayesian information criterion (BIC). The independent stochastic component was modelled using the normal distribution function. Overall, a very good agreement is observed between generated and historical records, with the Nash-Sutcliffe coefficient being equal to 0,98.