We analyse a unique data set that combines reservation wage and actually paid wage for a large sample of Dutch recent higher education graduates. On average, accepted wages are almost 8% higher than reservation wages, but there is no fixed proportionality. We find that the difference between reservation wage and accepted wage is virtually random, as search theory predicts. We also find that most information contained in the accepted wage is included in the reservation wage, as one would predict if individuals are well informed about the wage structure that characterizes their labour market.