Purpose The study aims to examine the export-led growth (ELG) hypothesis for Bangladesh. The direction of causality between export and output largely determines the success of export-oriented trade policies. A unidirectional causality running from export to output growth is required according to the narrow definition, while bidirectional causality is allowed for the broader definition. The study offers the causality inference, both from narrow and broader senses. Design/methodology/approach The study uses the bootstrap version of Toda and Yamamoto-modified causality tests, a recent development in time series econometrics, robust against the regularity conditions such as stationarity, properties of integration and cointegration and constancy of parameters. It uses monthly secondary data for the period of 1990-2014. Findings Test results suggest a unidirectional positive causal relationship from exports to output growth, meaning that the policies and strategies supporting exports are promoting output growth and thereby approve the ELG hypothesis for Bangladesh from the narrow sense. However, the absence of bidirectional causality between export and output growth, necessary to support the ELG hypothesis from the broader perspective, discards the conjecture that output growth is reinvigorated through the probable second-round effects of ELG produced from output growth to exports. Originality/value Beyond the conventional approaches, this is the first contemporary time series econometrics causality analysis between export and output growth of Bangladesh, both from narrow and broader senses.