This study aims to develop statistical models for forecasting the quantity of fish catches in the Songkhla Lake basin in southern Thailand. Data comprise a total monthly fish catch in tonnes from January 1977 to December 2006. We fitted an observation-driven model to the logarithm of the total monthly fish catch. The model contains seasonal effects and time-lagged terms for the preceding two months. We obtained an r-squared of 51% with both the seasonal and timelagged coefficients which was statistically significant. Although the catch has decreased substantially in the last ten years, no long-term trend is evident. This model can be used for short-term and possibly medium-term fish catch forecasting. The catch in the Songkhla Lake basin may have exceeded the sustainable capacity due to over-exploitation and illegal fishing. Strengthening the political will to develop enforceable and sustainable fishing practices is therefore desirable.