Exports of goods and services are a major driver of the economic and social development of countries, especially in the case of small open economies, such as Morocco. Since independence, Moroccan exports have been characterised by a concentration on basic products and a relatively weak evolution in comparison with competitors. Since 2005, these exports have undergone a profound structural transformation. The integration of the Moroccan economy in the new world trades has changed the base of specialisation of the country. It has gone from a country specialised in the production and export of agricultural products and textiles to a country producing and exporting automobile, electrical and aeronautical products. Therefore, forecasting Moroccan exports are of great importance for policy makers and government. The most popular and appropriate model for forecasting macroeconomic series in the literature are times series models. In this paper, we test whether the Auto Regressive Moving Average (ARMA) model can forecast Moroccan exports. To do so, we follow the modelling process developed by Box & Jenkins (1976), namely: identification of the susceptible models, parameter estimates and validation testing. The results show that satisfactory forecasts can be obtained using the ARIMA model (0,1,0) or AR (1). This confirms the idea that the behaviour of exports today is the result of a rich historical antecedent of events. The examination of the country's economic history allows us to deduce the importance of structural factors for the development of exports. This analysis can serve as the basis for a multivariate model for the behaviour of Moroccan exports in the future.