Empirical implementation of labour supply theories is usually based on actual hours of work. This requires strong assumptions on the impact of labour demand. To avoid these assumptions, subjective data on desired labour supply can be used. In this paper we investigate whether respondents' answers to survey questions on the desired number of working hours contain additional information on the respondents' preferences. Using panel data for the Netherlands, we analyse whether deviations between desired hours and actual hours of work help to predict future changes in the respondents' actual working hours. We use parametric and recently developed non-parametric tests. The results show that information on desired working hours is helpful in explaining female labour supply. For males the evidence is mixed.