Paper describes matching on Polish labour market comparing simultaneously the process mechanism – random versus stock – flow one. This aims at determining the role of stock and flow variables in generating outflows from unemployment. Analysed period, 1999−2010, reflects relatively consequent behaviour of the Beveridge curve. There are presented estimates of the matching function, also handling the bias resulting from data temporal aggregation. Such problem arises when discrete time data is used to describe the continuous-time process. Elements of the Markov transition matrix are exploited to approximate the magnitude of the effects corresponding to basic flows on the labour market influencing the matching process: on-the-job search, out-of- -labour force search and discouraged workers.