Despite the increasing empirical literature on foreign direct investment (FDI)-led growth, export-led growth (ELG) and import-led growth (ILG) hypotheses, this study investigates the validity of the FDI-export-growth and FDI-import-growth hypotheses in Turkey by using quarterly time series data for 1998:1-2009:1 period. To examine these linkages, we use the two-step procedure from the Engle and Granger model: In the first step, we define the order of integration in series and explore the long run relationships among the variables by using four unit root tests and autoregressive distributed lag (ARDL) bounds testing approach of cointegration test, respectively. In the second step, we test causal relationships by using the error-correction based causality models. The ARDL bounds test reveals that there is an evidence of a long-run relationship between the FDI, import (IM) and real gross domestic product (GDP), but no evidence of a long-run relationship between the FDI, export (EX) and GDP in Turkey. According to the causality test results for GDP-IM-FDI equation, there is evidence of two-way (bidirectional) causality between GDP and IM, evidence of one-way (unidirectional) causality from FDI to GDP and evidence of one-way Granger causality from FDI to IM. The existence of unidirectional causal links suggests that FDI strategies should be designed to promote economic growth.