This study uses the extreme bounds analysis of Leamer (1983) to identify some robust determinants of the long-run growth rate in seven South-Asian countries. The relationships between the two are estimated using panel data. We also consider some methodological issues concerning the specification. It is argued that the frequently used specification of the growth equation by the cross-country studies is inappropriate for estimating the long-run or steady state growth effects of variables such as the investment ratio. We use an alternative specification. Since the steady state growth rate in theoretical growth models depends on total factor productivity (TFP), we estimate the long-run growth effects of variables by analysing the determinants of TFP. This approach is suggested by a few influential economists and has been used by Senhadji (2000).