Hours worked is a time series of interest in many empirical investigations of the macroeconomy. Estimates of macro elasticities of labour supply, for example, build on this variable. Other empirical applications investigate the response of hours worked to a shock to technology on the basis of the real business cycle model. Irrespective of the problem being addressed, robust inference of empirical outcomes strongly hinges on the adequately modelling of the time series of hours worked. The aim of the present paper is to provide cross country evidence of the nonstationarity of hours worked for OECD countries. For these purposes, panel unit root tests are employed to improve power against univariate counterparts. Since cross section correlation is a distinct feature of the underlying panel data, results are based on various second generation panel unit root tests which account for cross section dependence among units. If an unobserved common factor model is assumed for generating the observations, there is indication for both a common factor and idiosyncratic components driving the non-stationarity of hours worked. In addition, taking these results together, there is no indication of cointegration among the individual time series of hours worked.