The problem of statistical calibration, when both standard and non-standard measurements are subject to error, is formulated as a predictive errors-in-variables model. Under the assumptions of unreplicated observations, a simulation study was conducted to compare relative performances of the ordinary least squares and the maximum likelihood estimation method, each combined with classical and inverse prediction. The Fuller method is also included for comparison. The mean squared error of prediction and the probability of concentration are adopted as criteria, based upon which guidelines are provided for selecting appropriate calibration procedures for the given situation.