We test whether the discouraged worker or added worker effect prevails in the Polish labour market. The discouraged worker effect implies that the participation rate is procyclical with respect to the GDP and countercyclical with respect to the unemployment rate. The added worker effect yields contrary findings. We analyse the period 1994–2014 with quarterly data. We focus on the working age population, both males and females. We apply a range of methods to obtain robust results, some of which have never been employed to resolve this problem. They include ad hoc filtration, spectral analysis, unobserved component model, time-varying parameter model, and frequency domain regression. The results indicate that the added worker effect prevails in most of the business cycle frequencies. It is significant and varies over time. It is true for both males and females. It is considerably stronger in contractions than in expansions. In low business cycle frequencies, the discouraged worker effect prevails. Although the last case is rare, it proves the heterogeneity of labour force behaviour over the business cycle.