Markovian models describe how the current measurement depends on the previously observed measures and on factors both observed and unobservable in the data. The dissertation underlines the several features that make Markovian modelling an interesting approach for life course studies focusing in particular on latent-based Markov models. In such an approach the dynamics of a characteristic of interest is explained by the evolution in time of a latent variable. This approach is particularly interesting in life course studies since many factors that are unobserved or difficult-to-observe in the data may influence a life history and can change over time. The dissertation discusses on a general framework for modelling longitudinal continuous data by the means of a latent based Markov model. The model incorporates a latent Markov process that governs the switching among regimes of change, the dependence between successive observations with an autoregressive component and the effect of the time-varying covariates. Empirical applications in population studies from three different sources are discussed.