An automated pavement inspection system consists of image acquisition and distress image processing. The former is accomplished with imaging sensors, such as video cameras and photomultiplier tubes. The latter includes distress detection, isolation, classification, evaluation, segmentation, and compression. We focus on wavelet-based distress detection, isolation, and evaluation. After a pavement image is decomposed into different-frequency subbands by the wavelet transform, distresses are transformed into high-amplitude wavelet coefficients and noise is transformed into low-amplitude wavelet coefficients, both in the high-frequency subbands, referred to as details. Background is transformed into wavelet coefficients in a low-frequency subband, referred to as approximation. First, several statistical criteria are developed for distress detection and isolation, which include the high-amplitude wavelet coefficient percentage (HAWCP), the high-frequency energy percentage (HFEP), and the standard deviation (STD). These criteria are tested on hundreds of pavement images differing by type, severity, and extent of distress. Experimental results demonstrate that the proposed criteria are reliable for distress detection and isolation and that real-time distress detection and screening is currently feasible. A norm for pavement distress quantification, which is defined as the product of HAWCP and HFEP, is also proposed. Experimental results show that the norm is a useful index for pavement distress evaluation.