This study presents a model capturing sources of Australian aggregate labour productivity using annual time series data from 1970 to 2001. Labour productivity, or real output per hour worked, in this model is determined by real net capital stock in information technology and telecommunications (ITT), real net capital stock in the non-ITT sector, trade openness, human capital, the wage rate, international competitiveness, and the union membership rate. Given the lack of long and consistent time series data, multivariate cointegration techniques are inappropriate as the cointegration results will be sensitive to the lag length, the inclusion or exclusion of the intercept term or a trend in the cointegration equation and/or the vector autoregression (VAR) specification. Therefore, the Engle-Granger representation theorem and the Hausman weak exogeneity test have been employed to determine the short and long-term drivers of Australian productivity. Empirical estimates indicate that, in the long-term, policies aimed at promoting various types of investment, trade openness, international competitiveness, and the use of wage as a stimulant in a decentralised wage negotiation system, will improve labour productivity. In the short term, all the above variables except for human capital and labour reforms, which both need more time to evolve, determine productivity performance.