a b s t r a c tLike other forms of mechanical damage, wear may progress in stages, the length and severity of which depend upon the type of wear and the nature of the tribosystem. Wear rates, wear coefficients, and wear factors are commonly reported as normalized quantities whose units imply a linear relationship with variables such as sliding distance, number of repetitive cycles, elapsed time, and normal force. Unfortunately, such implied linearity can be misleading in design-specific material selection. Examples of non-linear wear characteristics are provided for erosive wear, abrasive wear, fretting wear, and nonabrasive (also known as 'adhesive') wear. The differences between a system-specific, linearized wear rate and the instantaneous wear rate will be discussed, as will the difference between sequential and simultaneous wear transitions. The practice of linear normalization ignores such phenomena as incubation periods, running-in, and post-steady-state transitions, leading to misleading dependencies of wear rates on the sliding distance and sampling interval. Some current ASTM testing standards, notably those on erosion testing, account for the non-linear behavior of wear, but others, such as tests for sliding wear, usually do not. This article discusses non-linear wear behavior of several kinds, how the data can be treated, how wear rate notation can be improved, and how recognizing certain non-steadystate behavior can improve the basic understanding, modeling, and testing of wear.