Recovery from implantation of a cannula in the dorsal aorta (DA) of Atlantic salmon (Salmo salar) was studied in relation to nutritional status and sampling intensity. The incentive for the study was the inconsistency between published reports and our own experience of recovery and longevity of fish exposed to this protocol. In two studies using starved fish, blood (0.3 ml) was sampled 0, 1 and 24 h after DA-cannulation, and thereafter at 48 and 72 h and thereafter once weekly for four weeks. In a third study using fed fish, four consecutive samples (0, 3, 6 and 12 h after a meal) were obtained twice a week over a four-week period. All fish displayed a sharp increase in pCO(2) and haematocrit (Hct) during surgery, followed by a marked raise in cortisol, glucose, sodium and potassium (1 h). pCO(2), pH and Hct approached baseline levels as early as the 1 h post-surgery sample, while this was not the case for cortisol and electrolytes before the 24 h post-surgery sample. Glucose did not display any significant changes post surgery. From then on, all variables displayed minor but non-significant (P > 0.05) changes indicating a steady state close to baseline values for unstressed fish. This pattern was independent of sampling procedure, i.e. repeated single or multiple samples and thus volume of blood removed. Nutritional status (fed vs. starved) did not affect post-surgical recovery pattern. Only K(+) and Hct displayed consistent and significant post-prandial patterns. We found marked differences between baseline level of cannulated fish and uncannulated control fish, in pH, K(+) and Hct indicating that cannulation may be the preferred method to obtain representative resting values in fish.