This paper is devoted to the estimation of models of transitions between labour market states using data collected at discrete dates as is an ordinary longitudinal survey. The innovative feature of the paper is the use of ‘tenure' information in conjunction with point sample information on the state occupied to identify and efficiently estimate the parameters of a homogeneous Markov process. It is shown that using simultaneously tenure and state records allows full identification of the model and therefore that within- and between-status mobility can be distinguished. This dual type of information can be exploited to design a test of the Markov assumption. An empirical application on data drawn from a balanced male subsample of the 1986 to 1988 issues of the French Labour Force Survey is presented.