The main task of this paper is to conduct a comparative investigation of two very similar empirical studies, namely, Fosu (1990a) and Helleiner (1986), and to isolate factors (model specification, sample, period of estimation) that have led the authors to reach different conclusions concerning the role of the export sector in the process of growth. The paper then extends the Fosu and Helleiner tests to a later period. In order to explain the different conclusions, an extensive "sensitivity analysis" is conducted and we are led to conclude that different model specifications can best explain the difference in the authors' empirical results. The results indicate that measuring the export variable in models testing the export-led growth hypothesis is a crucial task, and that extended modelling must be part of the future research agenda. Copyright 1993 by Oxford University Press.