As one of the most wear monitoring indicator, dimensional feature of individual particles has been studied mostly focusing on off-line analytical ferrograph. Recent development in on-line wear monitoring with wear debris images shows that merely wear debris concentration has been extracted from on-line ferrograph images. It remains a bottleneck of obtaining the dimension of on-line particles due to the low resolution, high contamination and particle’s chain pattern of an on-line image sample. In this work, statistical dimension of wear debris in on-line ferrograph images is investigated. A two-step procedure is proposed as follows. First, an on-line ferrograph image is decomposed into four component images with different frequencies. By doing this, the size of each component image is reduced by one fourth, which will increase the efficiency of subsequent processing. The low-frequency image is used for extracting the area of wear debris, and the high-frequency image is adopted for extracting contour. Second, a statistical equivalent circle dimension is constructed by equaling the overall wear debris in the image into equivalent circles referring to the extracted total area and premeter of overall wear debris. The equivalent circle dimension, reflecting the statistical dimension of larger wear debris in an on-line image, is verified by manual measurement. Consequently, two preliminary applications are carried out in gasoline engine bench tests of durability and running-in. Evidently, the equivalent circle dimension, together with the previously developed concentration index, index of particle coverage area(IPCA), show good performances in characterizing engine wear conditions. The proposed dimensional indicator provides a new statistical feature of on-line wear particles for on-line wear monitoring. The new dimensional feature conveys profound information about wear severity. As one of the most wear monitoring indicator, the dimensional feature of individual particles has been studied mostly focusing on off-line analytical ferrograph. Recent development in on-line wear monitoring with wear debris images shows that merely wear debris concentration has been extracted from on- line ferrograph images. It remains a bottleneck of obtaining the dimension of on-line particles due to the low resolution, high contamination and particle’s chain pattern of an on-line image sample. In this work, statistical dimension of wear debris in on-line A two-step procedure is proposed as follows. First, an on-line ferrograph image is decomposed into four component images with different frequencies. By doing this, the size of each component image is reduced by one fourth, which will increase the efficiency of subsequent processing. The low-frequency image is used for extracting the area of ​​wear debris, and the high-frequency image is adopted for extra Second, a statistical equivalent circle dimension is constructed by equaling the overall wear debris in the image into equivalent circles referring to the extracted total area and premeter of overall wear debris. The equivalent circle dimension, reflecting the statistical dimension of larger wear debris In an on-line image, is verified by manual measurement. Two parameters are carried out in gasoline engine bench tests of durability and running-in. Evidently, the equivalent circle dimension, together with the previous developed concentration index, index of particle coverage area (IPCA), show good performances in characterizing engine wear conditions. The proposed dimensional indicator provides a new statistical feature of on-line wear particles for on-line wear monitoring. The new dimensional feature conveys profound information about wear severity.