The statistical technique for detecting outliers in multivariate non-Gaussian data on the basis of normalizing transformations and the Mahalanobis squared distance (MSD) is proposed. Application of the technique is considered for detecting outliers in the data of four-variate measurements.