We analyze the drivers of nonperforming loans in the Turkish banking system after the 2000-01 Turkish banking crisis. By constructing a vector autoregression model, we perform dynamic out-of-sample forecasts, which yield quite accurate results compared to the actual data. Since forecasting is a very crucial tool for both policy makers and market players, these results are some of the main strengths and contributions of this study. This article shows various patterns between the economic and financial indicators and the nonperforming loans. One important message obtained from the results is that policy makers should be concerned about the status of the economy and the market expectations to maintain stability in the banking system.