A data set from commercial Atlantic salmon (Salmo salar L.) producers on production intensity and production strategies in smolt tanks (N = 63-94) was obtained during 1999-2006. The effects of production intensity on subsequent fish mortality and growth during the early sea phase (90 days) were examined by principal component analysis and subsequent generalized linear model analysis. Levels of accumulated metabolites (CO(2), total ammonia nitrogen and NH(3)), and information provided by producers (production density (kg fish m(3-1)), specific water use (l kg fish(-1) min(-1)) and oxygen drop (mg l(-1)) from tank inlet to tank outlet), were used as predictor variables. In addition, several other welfare relevant variables such as disease history, temperature during freshwater and sea stage; season (S1) or off-season (S0) smolt production; and the use of seawater addition during the freshwater stage were analyzed. No strong intensity effects on mortality or growth were found. CO(2) levels alone (P < 0.001, R(2) = 0.16), and in combination with specific water use (R(2) = 0.20), had the strongest effect on mortality. In both cases, mortality decreased with increasing density. For growth, the intensity model with the most support (R(2) = 0.17) was O(2) drop, density and their interaction effects, resulting in the best growth at low and high intensity, and poorer growth at intermediate levels. Documented viral disease outbreaks (infectious pancreatic necrosis and two cases of pancreas disease) in the sea phase resulted in significantly higher mortalities at 90 days compared with undiagnosed smolt groups, although mortalities were highly variable in both categories. The temperature difference between the freshwater stage and seawater had a small, but significant, effect on growth with the best growth in groups stocked to warmer seawater (P = 0.04, R(2) = 0.06). S0 and S1 smolt groups did not differ significantly in growth, but the mortality was significantly (P = 0.02) higher in S1 groups. Seawater addition as a categorical variable had no significant effects, but when analyzed within the seawater addition group, intermediate salinities (15-25 ppt) gave the best results on growth (p = 0.04, R(2) = 0.15). Production intensity had small explanatory power on subsequent seawater performance in the analyzed smolt groups. If anything, the analysis shows a beneficial effect of intensive production strategies on subsequent performance. Analysis of the various production strategies indicates better survival of S0 compared with S1 smolt groups, improved growth when stocked in seawater warmer than freshwater, and a negative effect of viral disease outbreaks on survival. The results clearly demonstrate the difficulty of extrapolating results from experimental work on fish welfare and production intensity variables to commercial production. On the other hand, the presented results may simply demonstrate that the traditional fish welfare criteria growth and mortality may not suffice to evaluate welfare consequences of suboptimal water quality or production strategies in the aquaculture industry.