This paper models the evolution and determinants of the shares of agriculture, manufacturing and services to GDP for 4 South Asian countries (Bangladesh, India, Sri Lanka and Pakistan) for 55 years: 1960-2014. Determinants of these shares were classified into three broad categories “country fundamentals”, “policy” and “regional and reform dummies”. Several models including pooled OLS, GLS, panel and quantile regression are estimated. In general, the estimated models fit the data well. Policy conclusions regarding structural transformation are derived from the viewpoint of increasing the shares of the services and, particularly, the manufacturing sectors to GDP . We find that enhanced availability of electrical power and higher capital investment are central to the enhancement of the share of the manufacturing sector. The relationships of the shares with GDP per capita are fragile and, sometimes, counter-intuitive. It is a matter of concern that the impacts of key policy variables such as secondary school enrolment, FDI inflow and trade openness, are not robust across the models and in, some instances, have the “wrong” sign even when significant. It seems that South Asia has undergone a period of arrested industrial development. There is urgent need for policy intervention if this condition is to be redressed.