Norway farmed salmon annual deaths 54.8 million, two-thirds of causes attributed to injury, environment and "unknown causes"

The mortality data in the Norwegian Veterinary Institute's "2025 Fish Health Report" (Fiskehelserapporten 2025) gives a clear figure: 54.8 million salmon died in the seawater farming phase last year. Kjetil Korsnes, a fish health biologist and day-to-day manager of AquaPredict, wrote in the Norwegian aquaculture news website iLaks on 5 November 2026, raising a question about the report's classification of causes of death—about two-thirds of registered dead fish were caused by injury, environment or unknown causes. We know when and where fish are lost; what is missing is measurement of what happened inside the fish before death and why they were lost.

The report classifies the causes of death of registered dead fish as follows: injury and trauma 33.5%, unknown causes 27.4%, environmental conditions 4.3%, the three together accounting for about two-thirds. In the article, Korsnes says that "unknown cause" is an honest registration, but it is not a diagnosis; it means we have not measured the indicators that could have explained the cause. He notes that when handling operations (such as delousing) account for a large proportion of injuries, the question is rarely whether the fish was handled, but why it could not withstand it.

The article cites Øyvind Hilmarsen's recent view—that the aquaculture industry has data but does not use it—and expresses agreement, while raising another question: do we have the right data? Korsnes writes that production systems show when and where fish are lost, and environmental logs and water samples show what environment the fish are in, but neither shows what is happening inside the fish or whether the fish can compensate for the load it bears; this is precisely the missing link.

The method he gives is blood testing: taking a blood sample in a tank or net pen can quickly provide the needed information, simultaneously measuring key physiological responses such as osmoregulation (ion balance), respiration, stress load, kidney function and energy status, and comparing the measured values with weight-corrected normal ranges. The article says the new part is not the individual values, but reading these values as a mechanistic whole and combining them with water quality and production condition data from the same facility.

The article gives three facility cases using blood testing technology. First, in a hatchery facility, a mortality peak appeared the day after sorting, and the easy conclusion would be that sorting itself was the problem; but blood showed the fish had lost ion regulation capacity and could not recover, while a tank with the same water chemistry but no handling was normal; water quality analysis showed the cause—the chemical composition of the water was the difficulty, and after the facility added ions in salt form, mortality fell to background levels. Second, in another facility, the use of salt feed was tested before stocking, with two groups fed salt feed and ordinary feed; the salt group maintained ion balance and had lower mortality in the first week in seawater, but blood showed that in water with weak buffering capacity both groups bore the same load; salt feed helped the fish regulate ions, but the root cause was water chemistry, a common difficulty in many facilities; the water body sets a capacity ceiling, and blood shows whether and when this ceiling appears. Third, in a fish transport chain, blood was sampled before and after each handling step; ion loss began during transport, pumping at the receiving end was the heaviest step, and physiological indicators showed strong stress load and reduced osmoregulation capacity, thereby identifying the key step in the chain and setting targets for recovery before subsequent handling.

Korsnes summarizes the common pattern of these three cases: handling operations are often the trigger, while the environment determines whether the fish can withstand it; without blood data only the trigger is seen, and without water quality and production data the blood data cannot be fully understood; only the combination of the two gives the answer.

The article extends the same logic to the seawater phase: sample blood before delousing or other handling to see whether the fish can withstand it; sample blood immediately after handling to measure the load; sample blood in the following days to see whether the fish recover; and place these data alongside the net pen's temperature, oxygen, salinity, density, feeding and daily losses per pen. The article says that in this way part of the "unknown causes" can be converted into physiological explanations, handling dates can be scheduled according to the fish's condition rather than the calendar, and it can be judged before putting load on the fish whether it can withstand the planned delousing. The hatchery case also shows that fish may enter seawater with reduced osmoregulation capacity, bringing this problem into the seawater phase; if fish have underlying physiological difficulties, their robustness—the remaining capacity to withstand another load—will decline.

The article lists the data categories required: production data, i.e. handling logs linked to density and daily losses per unit; environmental measurements, i.e. data indicating where and when measurements were taken and showing the load on the fish; blood samples, i.e. data collected before (and after) key operations and showing the state the fish are in. The article concludes by writing: measure, understand, take action, measure the effect again; the fish have always been showing us what is wrong with them, we just have not measured it; only in this way can the mortality rate of the farming industry be reduced, getting closer to the target of about 5%.

The sources marked in the article include: Modal, T., Wiik-Nielsen, J., Oliveira, V.H.S., Svendsen, J. C. og Sommerset, I. 2026. Fiskehelserapporten 2025, Chapter 2 (mortality figures) and Chapter 2.5 (classification by cause, Table 2.5.2), Norwegian Veterinary Institute Report Series No. 5a/2026; the iLaks article "The aquaculture industry has the data, it is time to use them" (Øyvind Hilmarsen, FHF); and data cases from three facilities using blood testing technology provided by AquaPredict.