In this paper, a continuous version of the Markov Chain Model (MCM) is proposed to project the number and the population structure of farms. It is then applied to the population of professional French farms. Rather than working directly with transition probabilities as in the traditional, discontinuous, MCM, this approach relies on the close but not identical concept of growth rate probabilities and exploits the Gibrat’s law of proportionate effects which appears to be supported by the French data. It is shown that the proposed continuous MCM is a more general approach, since it enables to derive more in-depth detail on the distribution of the projected population and the traditional MCM transition probability matrix can be easily reconstructed from the estimated growth rate probabilities. Though the continuous MCM is presented in this paper in a stationary framework, it should be possible to develop a non-stationary version in a similar way traditional MCMs are now made non-stationary.