In this paper, models are developed to explain and forecast the effect of world cocoa price on production of cocoa in Ghana by using regression model with time series errors. The focus of the investigation was to find out whether the world cocoa price can assist in forecasting the future behavior of the cocoa production in Ghana. Annual data from 1961 to 2010 were used in fitting the model while 2011 and 2012 were used as out-of-sample data. Based on the behavior of several model adequacy techniques, the regression model with ARIMA(2,2,0) errors was considered as the ‘best' model for the production variable. The mean absolute percentage error (MAPE), as a forecast accuracy measure, was used to validate the model. Thus, the MAPE of the regression model with ARIMA (2,2,0) errors was 7.97%. However, the conventional ‘best' ARIMA model fitted to the production variable indicated an MAPE of 16%. This shows that, the production variable has smaller MAPE, when it was modeled together with world price using regression with ARIMA errors. Hence, regression model with ARIMA (2,2,0) errors is a better statistical technique in forecasting production of cocoa in Ghana than the conventional ARIMA method.