The effect of unionisation on productivity is an unresolved empirical question. Many studies have found a range of results. Unlike nearly all preceding studies, this study is based on a sharply defined natural experiment which is provided by the Australian Workplace Relations Act 1996. This Act increased flexibility in bargaining at unionised firms as well as directly and indirectly reducing union bargaining power. Using a sample of union and non-union small and medium sized firms between 1994/95 and 1997/98 we estimate, using the Levinsohn-Petrin methodology, Cobb- Douglas production functions and total factor productivity. Identification is enabled by the use of stock rather than flow measures of labour. We find relatively rapid growth in unionised firm productivity growth following the introduction of the WRA consistent with reducing trade union power leading to an increase in productivity. This increase, though, is typically not statistically significantly different from zero.