This paper develops a theoretical model of the simultaneous determination of union wages and union membership, and empirically implements the model using the 1990 Workplace Industrial Relations Survey. The empirical literature on union wage gaps has long recognized that union membership may be endogenous in equations estimating the determinants of union wages. Although some empirical studies have effectively endogenized membership to control for simultaneity, the estimated relationships have generally followed an ad hoc approach in specifying the union membership equation. Indeed, the empirical literature on union wage gaps in a simultaneous equation framework, and the small theoretical literature examining unionization, have largely proceeded separately, with so far little interaction between the two. This paper aims to rectify this imbalance. The innovation of the paper is that it provides a formal model of union wage and membership determination where the union is recognized for bargaining purposes, where membership of the union is not compulsory, and where the union faces a budget constraint in its organization activities. This simultaneous equation model is then used as the basis for estimation of manual wages and union density for the private sector. The results reveal that for our particular data set, union density significantly increases the hourly manual wage rate, and that union wages have a significant impact on the membership decision. Moreover, other union-provided goods (apart from the wage rate) affect union membership, suggesting that in the absence of coercion, unions may still retain some membership providing they can continue to provide non-wage benefits which accrue exclusively to their membership.