Using data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey, our research indicates that unobserved heterogeneity substantially biases cross-sectional estimates of union wage effects upward for both males and females. Estimates of the union wage premium for male workers between the ages of 25 and 64 fall from 8.7 percent to 5.2 percent after controlling for unobserved heterogeneity. For females aged 25 to 63 the estimated 4.0 percent cross-sectional union wage premium falls to 1.9 once unobserved heterogeneity is controlled for. Our results also indicate positive sorting by unobserved skills into union membership, especially among low skilled male and female workers. There is also evidence of negative sorting into unions among the most highly skilled.