Regression analysis using level variables is a popular technique for identifying the demand-side determinants of foreign direct investment. While the technique is justifiable in analyses with cross-country data, it is not with time-series data from a single country. The lack of a logical relationship between the dependent and explanatory variables in an analysis using time-series data usually leads to spurious and misleading results. With an alternative specification, FDI data of Taiwan were utilized to highlight this problem. Contrary to conventional wisdom, FDI in Taiwan was found to be probably supply-side determined.