Fish welfare is an area of increasing interest and stocking densities used in commercial aquaculture have been highlighted as a subject of specific welfare concern. The present study assessed how stocking density affects growth performance (final body weight, SGR, FCR), physiological (plasma cortisol, glucose, lactate) and behavioural (swimming activity, stereotypes, escapes, aggression, airbreathing) responses of farmed African catfish (Clarias gariepinus). We stocked African catfish (10g) in triplicate in 120l tanks at densities of 500, 1125, 1750, 2375, and 3000animals/m3. After exposure to stocking density, responses (plasma cortisol, glucose, lactate, number of lesions) to an acute stress challenge were studied. Growth performance, physiological and behavioural data showed little indications of impaired African catfish welfare within the range of densities examined in this study. However, the acute stress challenge gave indications that African catfish were in fact stressed. Catfish housed at densities of 500 and 3000fish/m3 showed signs of chronic stress, reflected by the absence of a cortisol response after an acute net stressor. Furthermore, fish housed at high densities (2375 and 3000fish/m3) showed a strong increase in the amount of skin lesions after the 1-h group housing after net-stress. In conclusion, an interaction effect between stocking density and additional stress was shown. Since additional stressors (e.g. handling, grading) are not uncommon in a farming situation, this means that the total farming situation may influence the effect of stocking density. Welfare of African catfish seemed impaired at both the lowest and highest density. Due to opposing data, the situation of fish housed at densities between 1125 and 2375fish/m3 is less clear. The higher aggression at the low end of this density range and increased escape attempts at the high end of this density range seem to be balancing each other. Which stocking density within this range should be regarded optimal depends on the relative weight assigned to the different parameters.