The present paper uses a combination of workplace and linked employee-workplace data from the 1998 Workplace Employee Relations Survey and the 2004 Workplace Employment Relations Survey to examine the impact of unions on training incidence, training intensity/coverage, and training duration. It also examines the impact of unions and training on earnings and a measure of establishment labour productivity. In addition, the implications of training for the firm’s bottom line are evaluated. Union effects on training emerge as fairly subtle, and are more positive when using individual rather than plant-wide training data. A positive impact of training on earnings is detected in both the individual and plant-wide wage data, albeit only for the earlier survey. Consistent with other recent findings, the effects of union recognition on earnings are today rather muted, while union-training interaction effects vary greatly. Instrumenting training provides positive results for the labour productivity outcome and, in the case of the earlier survey, for the financial performance indicator as well. However, some negative effects of unions are now also detected.