Grading of cultured flatfish is essential to maintain even sizes and to avoid potential feeding dominance or cannibalism. Current hand-grading and forced mechanical grading methods result in labour costs and additional stress to animals, reducing their growth performance. This study tested a self-grading system allowing turbot (Scophthalmus maximus) to self-grade, based on size, between tank sections within 8000 L tanks designated as “Graded (G)” and “Non-Graded (NG)”. The grading success (percentage of graded animals/total gradable individuals) with stimuli (light, feeding sequence and colour of the grading device) and without stimuli was analyzed over 9 days. Mean grading success without stimuli was 38.1 ± 11.5%. Grading success was not improved by changes in feeding sequence (30.6 ± 10.2%) or altered colour of the grading device (30.9 ± 13.8%). Marked increases in grading success were obtained by adding a light source to the G section (52.9 ± 12.4%), and by combining the presence of a light source and the feeding sequence (57.9 ± 14.4%). The combination of light sources and changes to the colour of the self-grading device resulted in the highest mean grading success (72.1 ± 18.2%) overall. Light and colour stimuli combined yielded the highest self-grading maximum (88.36 ± 5.3%), achieved on day 6 after experimental onset. A three-parameter asymptotic exponential equation was fitted to the data from each stimulus and parameters describing the curves of all conditions are reported. The present results indicate that the tested self-grading systems for turbot can be highly efficient if used with the appropriate stimuli. Self-grading systems operating at the measured levels of success offer a viable alternative to stressful grading events, avoiding significant commercial disadvantages as well as improving animal welfare and performance.