Atlantic salmon juveniles were subjected to a precocious (day 104 after first feeding) photoperiodic (P) and thermic (T) control. After rearing the fry at 16 °C, constant (16 hours of light and 8 hours of darkness, 16 °C) (C) and shortened simulated natural S photoperiod and temperature regimes were used in a factorial experimental design until transfer to seawater. The growth obtained with constant (C) regimes was better than with S regimes. In all cases, we observed very early (day 141, mid-June) a clear segregation of the population into two different subpopulations. Among the conditions, all the differences between the upper and lower modes were growth-dependent. We observed different developmental trends of the gill (Na+-K+)-ATPase activity: the S regime gave a very progressive increase of the enzymatic activity, while we had a rapid decrease after a first phase of increase with a constant photoperiod. The absolute levels of the gill (Na+-K+)-ATPase activity were slightly lower than those observed in one year-old smolts. However, they are quite similar to those of upper-mode fish that have a good salinity tolerance. After transfer to seawater, fish subjected to shortened simulated temperature and photoperiod had the lowest mortality and osmotic disturbance (after 48 hours and 3 weeks). Although the freshwater phase generally lasts more than 15–17 months for this species in a natural environment, we concluded that it is possible to obtain a good salinity tolerance after 7–8 months, with shortened simulated natural photoperiod and temperature regimes, in spite of the absence of a real smoltification; these regimes, however, must be applied during the "dynamic" presmolting phase when the growth is intense.