We discuss two approaches to multilateration with a particular focus on numerical aspects for a given dataset. More precisely, we calculate the volumetric errors of the MCV 754 milling machine from the data measured by a LaserTRACER and provide criteria showing which numerical method is appropriate for the solution of the multilateration problem with respect to a given measurement dataset. In the first case, we consider two cost functions of the optimisation problem; in the second, we use the extremal principle method. We discuss the calculation accuracy depending on the matrix condition number. Our results are compared to the reference results produced by the Trac-Cal software, which is a standard used by most producers for error compensations.