Stress in organisms results in energy dissipation, making developmental pathways less stable. Effects of chronic stress, manifested as small random departures from phenotypic symmetry, reflect developmental instability, are considered to be epigenetic and an effect produced by compromised fitness. Instability is detectable and effectively interpreted among sites or populations if samples are collected randomly, the stressor is present throughout character development, characters are identified accurately and excessive mortality does not erase the existence of developmental instability. Bacterial kidney disease (BKD) is a chronic systemic disease in salmonids that, after vertical transmission from parent to egg, persists and spreads throughout ontogeny, potentially affecting developmental processes. Because levels of progeny infection reflect parental infection levels, groups of offspring from parents with high and low levels of BKD infection can be compared to assess the effects of disease-mediated developmental stress. Analyses of fluctuating asymmetry in five bilateral characters were inconclusive, but significant reductions in the proportion of unusable scales, in the number of circulus errors, and in the directional asymmetry of branchiostegal rays were observed in fish from the high-BKD group. This group also contained individuals of significantly larger size. These results are opposite to those expected from traditional developmental instability theory in suggesting that surviving high-BKD fish have greater developmental stability. This reversal appears to be produced by selective mortality having a greater effect than sublethal stress in altering developmental instability patterns. These results are discussed with respect to size selectivity, heterosis and the assumptions supporting developmental instability as a tool for detecting chronic sublethal stress.