An automated fish size classifier with machine vision was developed. The knowledge base used by the system was obtained from experiments involving the actual physical measurement of tuna fish using camera-based image processing. The classification of the images of the fish into small, medium, and large was done according to a rule-base. The fish size classification used a look-up table that stored the size-image equivalence (pixels) and converted the size of fish into weights. The sorted fish were transferred into bins by using a flipper and stored according to size. A controller detected the presence of the fish on top of the conveyor and automatically stopped the latter's motion when no more fish were available. The results of the physical experiments showed that the controller developed in the laboratory was efficient, robust and adaptive.