Available statistics on growth trends in the Economic Community of West African States (ECOWAS) are wanting, particularly net per capita growth rates. The analysis of available data from 1970 to 2012 by this study, for instance, shows that the net real GDP growth rate for the ECOWAS is 0.52%. Only four countries had net growth rates above 1% per annum mean growth rate of ECOWAS region. At the estimated growth rate, the prospect of accelerated growth in ECOWAS is very weak. The Barro endogenous growth model states that government provision of services can generate externalities to the private productive activities. Government’s provision of productive services in ECOWAS can ensure long-run per capita output growth without the per capita growth rate running into steady state growth. However, there are divergent views as to whether government provision of services induces long run economic growth. These views are based on different schools of thought. For instance, the economic freedom school argues for minimum government involvement (small governments) to ensure economic and political freedom to induce private investors invest and encourage economic growth. The optimal government school of thought (medium size governments) argues that government spending enhances private productivity growth through the provision of infrastructure, spending on research and development, public education, sewage, other public goods and protection through functional law and order systems. The optimal school of thought also acknowledges that government expenditure can also reduce economic growth through increases in taxation. An increase in taxation reduces the returns on investment of physical and human capital and in research and development (R&D) of private firms. This thesis investigates the impact of government expenditure on the provision of public services on economic growth in ECOWAS. To assess the impact of government expenditure on the provision of services on economic growth of ECOWAS, this thesis assesses whether the size of government, government expenditure and economic institutions promoted economic growth in ECOWAS. The thesis also determines whether per capita government capital expenditure, per capita government consumption expenditure, per capita private capital stock, per capita manufacturing output, per capita services output and per capita agricultural output have any impact on per capita real GDP growth in ECOWAS. To carry out this study, data were collected from United Nations Conference on Trade and Development (UNCTAD) database and Transparency International (TI) database. The data used covered the period of 1970 – 2013. The statistical research methods applied are the time-series methods of panel unit root test, panel co-integration test, and panel regression analysis, using both panel OLS regression models and estimation and inferences in co-integrated panel data regression methods. The panel OLS regression models applied are the panel OLS regression; panel fixed effect model (FEM) regression and the panel random effect model (REM) regression. The estimation and inferences in co-integrated panel data regression models applied are panel VEC regression model, panel DOLS regression and panel FMOLS regression. The panel DOLS regression and panel FMOLS regression models do not have an intercept, unlike their pure time-series models, which have intercepted. To ensure that the parameters estimated are reliable, this thesis conducted diagnostic tests to subject the regression result to scrutiny. The estimated panel data regression using panel OLS regression, panel FEM regression and panel REM regression indicate that the results of the estimated parameters were spurious having both autocorrelations and heteroscedasticity. High values of adjusted R-squares that were approaching one and high significant values of t statistics but very low values of Durbin-Watson Statistics demonstrated the existence of heteroscedasticity and autocorrelation in residuals. The results of the diagnostic tests also show that the DOLS estimated regression model out-performed both VEC and FMOLS regression models based on both aggregate data and per capita data estimated parameters. The results of the parameter estimated using panel VEC and panel FMOLS regression models showed that both panel VEC and panel FMOLS regression models had the problems of their residuals having not only autocorrelations but heteroscedasticity. The panel DOLS regression results were satisfactory, having no multicollinearity, autocorrelations and heteroscedasticity. The estimated panel DOLS regression results were applied to test hypotheses formulated to guide this thesis. Results from panel DOLS estimated parameters show that the existing government size in ECOWAS stimulated economic growth. The results also showed that the government expenditure exhibited an inverted U-shape with respect to economic growth. The thesis also showed that existing government size in ECOWAS significantly stimulated economic growth in the region. The results of regression indicate that economic institutions contribute negatively to the economic growth of the ECOWAS. The results also established that government capital expenditure per capita has significantly engendered economic growth. Government consumption expenditure per capita stimulated economic growth. However, private capital stock per capita has not stimulated economic growth in ECOWAS. Service sector output per capita, agricultural output per capita and manufacturing output per capita stimulated significantly economic growth in the ECOWAS sub-region.