The knowledge of skilled cooks at the Shimonoseki fish market was analyzed to develop a quality evaluation system for dressed Purple puffer (Takifugu porphyreus) Migaki. To determine what aspects of fish surface color, freshness and texture are used by cooks for the evaluation its quality, a total of 145 fish were dressed and graded into five classes by the cooks. Then, the color and freshness of the meat were investigated every 24 hours, at 0-, 24, 48, and 72-hour time points. Analysis of the cooks' visual evaluation in comparison with the recorded information on fish color and meat freshness indicated that the cooks' evaluations accurately corresponded to the actual freshness of the meat. A statistical model was then constructed and evaluated using the colors recorded at these time points as four explanatory variables； the results corresponded to the assessments made by the cooks with 60% certainty. These results confirmed the validity of the quality evaluation system proposed here, which incorporates the knowledge of skilled cooks into its design for evaluating the quality of dressed purple puffer.