The use of statistical methodology for analytical methods validation is vital to assure that measurements have the quality level required by the goal to be attained. This thesis describes a statistical modelling approach for combining four different aspects of validation: checking the linearity of the calibration curve and compute the detection and the quantification limits; controlling the specificity of the analytical method; estimating the accuracy (trueness and precision) of the alternative method, for comparison with a reference method. The general approach is a synthesis of several partial techniques found in the literature, according to a choice of the most appropriate techniques in each case. For determination of the response function, statistical techniques are used for assessing the fitness of the regression model and for determination of the detection limit and the quantification limit. Method specificity is evaluated by adjusting a straight line between added and recovered concentrations via least squares regression and hypotheses tests on the slope and intercept. To compare a method B with a reference method A, the precision and accuracy of method B are estimated. A 4-factor nested design is employed for this purpose. The calculation of different variance estimates from the experimental data is carried out by ANOVA. The Satterthwaite approximation is used to determine the number of degrees of freedom associated with the variance components. The application of the methodology is thoroughly illustrated with step-by-step examples.