The objective of this study is to analyse further the governance–performance relationship while improving on two methodological issues: control for endogeneity and firm performance measurement. To mitigate the endogeneity problem, we first focus on subsamples of firms for which we ex ante expect better corporate governance to cause better performance. Second, we use generalized least squares regressions for panel data. To control for potential measurement bias, we measure firm performance using data envelopment analysis (DEA). The research is conducted in Canada over a five-year period from 2001 to 2005. Corporate governance is measured based on the Report on Business corporate governance index published by the Globe and Mail. Overall, the results show that better governed firms are more efficient. This study is in line with a growing number of recent studies that propose alternative measures of firm performance. By using DEA, this study brings together the corporate finance and productivity literature.