At each point P referring to the vertical datum i, an observation equation like Eq. [2] can be formulated. At those points referring to two neighbouring vertical datums i, i+1 (see Fig. 1), the observation equation takes the form: Eq. [4] Least-squares estimation of the vertical datum parameters Equations [2] and [4] can be solved by a least-squares adjustment. The functional and stochastic models are given by: Eq. [5] Eq. [6] A is the design matrix containing the coefficients of the unknowns in the observation equations; x is the vector of the unknowns; v contains the residuals; l contains the left-hand side elements in Eqs. [2] and [4]; is the expectation operator; C represents the variance-covariance matrices of the input data. The least-squares solution provides estimates for the vertical datum parameters W 0i