Shortcomings in the treatment of intangible investment in company accounts imply that there is no statistical collection for innovative activity which abides by the logic used for other economic activity data. As a consequence, analysts rely on innovation proxies derived from administrative and survey data. However, it is still unclear exactly how the different proxies are correlated, and whether the choice amongst different proxies matters. In the light of the innovation measurement, this paper takes another look at the relationship between different proxies of firm innovation. The results show that firm-level correlations between survey-based indicators and other proxies for innovation are highest for manufacturing firms and for product innovations. Copyright 2009 The Authors. Journal compilation 2009 Blackwell Publishing Ltd/University of Adelaide and Flinders University.