Abrasive flow machining (AFM) is very effective and widely adopted superfinishing process of internal channel surfaces in industry. There have been high demands for process monitoring of surface roughness evolution during AFM, as the evolution of surface roughness is sensitive to AFM media variables, such as abrasive grain size and concentration, as well as process duration. Acoustic emission (AE) is known to be a promising tool to detect microscale deformation mechanisms arising from abrasion. This work has shown there is a close correlation, during AFM with different media, between the evolution of surface roughness and material removal with the AE root-mean-square (RMS) and AE fast fourier transform (FFT) signals. Moreover, AE signals are correlated to wear mechanisms, such as ploughing and cutting mechanisms.