The objective of this paper is to examine the causality among exports, imports and income in Bangladesh. Bangladesh provides a good case for studying this relationship because of significant dependence of its major export, namely, ready made garments, on imported inputs. This paper differs from the existing studies in several ways. First, the study uses a long data set covering the period 1973-2008. Second, it takes into account various modeling issues that arise in causality framework and overcomes the omitted variable bias. It employs augmented Dicky Fuller and Phillips-Perron tests to examine the time series properties of exports, imports and income. Further, Johansen and Juselius test is used to examine the cointegration properties of the variables. Finally, the study examines both short-term and long-term dynamic relationships among the relevant variables within an error-correction framework.