This paper analyses the effects in the long-run between tourism and the economic growth in the regions of Antioquia, Bolivar, Bogotá, Magdalena and San Andrés and Providencia of Colombia. Using annual data from 1990 to 2005, the study uses cointegration analysis to consider the existence of Vector Error Correction Model (VEC) among real per capita Gross Domestic Product (GDP), tourism expenditures and real exchange rates. We show that the causality relationship is positive and unidirectional for all the regions but the values of elasticity are considerable different. Finally, we compare our study with similar papers also investigating the tourism-led growth hypothesis.