This research adds new insights to the analysis of the role trade unions play in the determination of wage and employment levels. It first identifies the puzzles faced by economists in the mid-nineties. Picking up one of these issues, it is here argued that the extent of wage rigidity and employment adjustment to shocks also depends on the degree of uncertainty faced by agents. Sequential bargaining models are proposed as the best instrument to be used in empirical research, as they nest other specifications and allow for efficient and inefficient outcomes depending on union power and on the degree of uncertainty faced by agents at the bargaining table. The empirical research carried out for Uruguay takes advantage of the unique situation of having temporal data from 1975 to 1999 for the same economic sectors with and without unions. Further, during the time in which unions were active two different bargaining structures were observed. The empirical results show that the re-appearance of trade unions in the eighties generated inflexibility in the labour market as bargaining was done at the industry level in a co-ordinated way. Wages were set above their market clearing levels at the expense of lower employment that was unilaterally set by management. However, unions were also able to reduce output and wage elasticities of labour demand relative to the seventies, when there was a ban on unions. In the nineties, bargaining over employment and decentralised negotiations promoted lower wage increases than in the previous decade and even wage cuts. Union action also reduced the proportion of non-production workers used by firms so that they promoted that management moved towards more skilled labour intensive technologies. However, they granted job stability by buffering the negative effects of external shocks and demand fluctuations both on employment levels and its composition.