New statistical characteristics are suggested for analyzing nonwoven fabric irregularity. These characteristics are based on the concept of outliers of random field irregularity beyond (or beneath) a fixed level, and allow the area and volume of these outliers to be determined. Outlier statistics are studied and results are presented for calculating their confidence intervals and the dimensions of the sample. The results of the simulation are presented, and the influence of nonhomogeneous random field irregularity on the outliers is depicted. The experimental part of the paper presents the results of a study in micro- and macroscale of some nonwoven fabrics. Some practical aspects of the new characteristics of nonwoven irregularity are also given. Typical application problems are set up, and examples of solutions are given.