How Are Fish Killed for Food? The Welfare Footprint of Slaughter
Summary
Most fish are killed by air asphyxiation or ice slurry — slow methods that, according to a 2025 peer-reviewed study (Schuck-Paim et al., Scientific Reports), cause rainbow trout roughly 10 minutes of moderate-to-extreme pain per fish, with time to loss of consciousness ranging from 2 to 25 minutes. Around 124 billion farmed fish and up to 2.3 trillion wild-caught fish are killed each year, and fewer than 1 percent are covered by species-specific welfare law at slaughter. Electrical or percussive stunning can render fish insensible in 1-2 seconds and is one of the most cost-effective animal-welfare interventions known, but it is not yet standard practice in most of the global industry.
Supported by 5 cited sources
Key Points
- 1The dominant commercial slaughter methods for fish are air asphyxiation (removing fish from water and letting them suffocate), ice slurry or 'live chilling' (immersion in near-freezing water), and CO2 narcosis. All three are slow. Fish gill lamellae stick together in air, drastically reducing oxygen exchange, and fish are cold-blooded — their brains tolerate hypoxia far longer than mammalian brains. Ice slurry can extend rather than shorten suffering, because cold-tolerant species remain conscious while their metabolism slows.
- 2A 2025 peer-reviewed analysis (Schuck-Paim et al., Scientific Reports, June 2025) is the first to put rigorous numbers on this. Using the Welfare Footprint Framework with behavioral and neurophysiological proxies for consciousness, the authors estimate that rainbow trout slaughtered by air asphyxiation experience a median of approximately 10 minutes of moderate-to-extreme pain per fish (95 percent credible interval 1.9 to 21.7 minutes), or roughly 24 minutes per kilogram of fish. Time to loss of consciousness ranged from 2 to 25 minutes depending on body size and water temperature.
- 3The study is co-authored by Lynne Sneddon, the foremost neuroscientist working on fish pain, and is published in a high-impact, peer-reviewed venue. It is single-species (rainbow trout) but methodologically transparent and directly policy-relevant.
- 4The scale is staggering. Mood et al. (2023, Animal Welfare) estimate that approximately 124 billion farmed finfish were slaughtered for food globally in 2019 (range 78 to 171 billion), about nine times the 1990 figure. Wild-caught fish killed annually are estimated at 0.78 to 2.3 trillion (fishcount.org.uk). Across both categories, fish are by a very wide margin the most numerous vertebrates humans kill for food.
- 5The legal protection gap is severe. Mood et al. (2023) report that countries representing roughly 70 percent of global farmed-fish production volume have no legal welfare protection at slaughter, and fewer than 1 percent of farmed fish (by count) are covered by species-specific welfare law. Most national slaughter regulations were written for terrestrial mammals and either exclude fish entirely or apply only generic standards.
- 6Welfare ranking of slaughter methods, from best to worst on the available evidence: percussive (priest) stunning to the head, under 1 second when accurately delivered; electrical stunning, 1 to 2 seconds to insensibility when current parameters are correct; CO2 narcosis, several minutes of distress before loss of consciousness; ice slurry / live chilling, 2 to 25-plus minutes; air asphyxiation, typically 2 to 25 minutes (with reports of up to 250 minutes in some cold-tolerant species); salt baths and live evisceration, severe and prolonged. Electrical and percussive stunning are the only methods that reliably produce rapid insensibility.
- 7Stunning is also one of the most cost-effective animal-welfare interventions known. Schuck-Paim et al. (2025) estimate that electrical stunning could avert 60 to 1,200 minutes of moderate-to-extreme pain per US dollar of capital expenditure, depending on the system and farm context. Despite this, most global production still relies on asphyxiation or ice slurry.
- 8The precautionary case is strong. Even those biologists who remain skeptical of mammalian-style conscious pain in fish (Rose 2002, Key 2016, Diggles and colleagues) accept that fish have functioning nociceptors. Sneddon (2019, Phil. Trans. R. Soc. B) and the 2018 Animal Sentience commentary symposium synthesize multi-decade behavioral and neurophysiological evidence supporting pain capacity. The 2024 New York Declaration on Animal Consciousness, signed by 500+ scientists, judges conscious experience in all vertebrates including fish 'at least a realistic possibility'. Given that, multi-minute slaughter durations for trillions of animals each year is a welfare problem of the largest possible magnitude.
Evidence Summary
The methodological centerpiece is Schuck-Paim et al. (2025, Scientific Reports). The authors apply the Welfare Footprint Framework — a system that combines behavioral indicators (gill flaring, escape attempts, vestibulo-ocular reflex loss, cessation of opercular movement) and neurophysiological proxies (EEG suppression, evoked-potential loss) to estimate the duration and intensity of negative affective states during slaughter.
Several caveats apply. First, pain duration in Schuck-Paim et al. (2025) is inferred from behavioral and neurophysiological proxies rather than direct subjective report — as in all non-human pain research. Second, the 2025 study is single-species (rainbow trout); extrapolation to other commercial species is plausible (trout are physiologically representative of cold-water salmonids and similar to many farmed finfish) but unreplicated. Third, a minority of biologists — notably Rose (2002), Key (2016), and Diggles and colleagues — continue to dispute that fish have conscious pain experience, arguing that fish lack the neocortical structures that mediate pain in mammals; the majority response is that convergent evolution can produce functionally equivalent experience through different neural architectures, but the disagreement is real and unresolved. Fourth, some operations DO use stunning — notably parts of the EU farmed-salmon and trout sector, which have moved toward electrical stunning under pressure from welfare regulators and certifiers — so 'fish are not stunned' is a generalization, not a universal. Finally, capital-expenditure cost-effectiveness numbers depend on plant context, throughput, and current method displaced.
Supporting Evidence
Schuck-Paim et al. (2025, Scientific Reports) applied the Welfare Footprint Framework — combining behavioral indicators (loss of vestibulo-ocular reflex, cessation of opercular movement, escape attempts, gill flaring) and neurophysiological proxies (EEG suppression, evoked-potential loss) — to estimate the duration and intensity of negative affective states during commercial air-asphyxia slaughter of rainbow trout. The headline result is a median of approximately 10 minutes of moderate-to-extreme pain per fish, with a 95 percent credible interval of 1.9 to 21.7 minutes, or roughly 24 minutes per kilogram. The paper is co-authored by Lynne Sneddon, the leading neuroscientist working on fish pain.
Caveats: Pain duration is inferred from behavioral and neurophysiological proxies rather than direct subjective report, as in all non-human pain research. The study is single-species (rainbow trout); extrapolation to other commercial species is plausible but unreplicated. A minority of biologists (Rose 2002, Key 2016, Diggles) continue to dispute conscious pain experience in fish.
Schuck-Paim et al. (2025) report time-to-loss-of-consciousness ranges of 2-25 minutes for rainbow trout under air asphyxia, varying with body size and water temperature. Sneddon (2019, Phil. Trans. R. Soc. B) synthesizes the broader nociception and slaughter literature, including evidence that fish gill lamellae stick together in air, drastically reducing oxygen exchange surface area, while fish cold-blooded metabolism allows the brain to tolerate hypoxia far longer than in mammals. Cold tolerance in many farmed species (carp, trout, salmon) means ice slurry slows metabolism while consciousness persists, extending the period of suffering rather than shortening it. Some species can take up to roughly 250 minutes to lose consciousness under air-asphyxia conditions.
Caveats: Loss-of-consciousness endpoints are inferred from behavioral and neurophysiological criteria; precise transition timing is necessarily probabilistic. Species, size, and water-temperature dependence means individual experiences vary widely.
Mood et al. (2023, Animal Welfare) used FAO production tonnage and species-specific mean-weight data to estimate that approximately 124 billion farmed finfish (range 78 to 171 billion) were slaughtered for food globally in 2019 — about nine times the 1990 figure. Wild-caught individual-fish estimates from fishcount.org.uk (Mood and Brooke), using a similar tonnage-to-individuals conversion across reported global landings and bycatch, yield a range of 0.78 to 2.3 trillion wild fish killed annually. Combined, fish are by a very large margin the most numerous vertebrates humans kill for food.
Caveats: Estimates depend on FAO tonnage data (which has known underreporting issues) and on species-specific mean-weight assumptions; ranges reflect this uncertainty. Bycatch and discards add further numbers that are even harder to quantify.
Mood et al. (2023, Animal Welfare) reviewed slaughter regulations across the major fish-producing jurisdictions and found that countries representing roughly 70 percent of global farmed-fish production volume have no legal welfare protection at slaughter, and fewer than 1 percent of farmed fish (by count) are covered by species-specific welfare law. Most existing animal-welfare slaughter regulations were drafted for terrestrial mammals and either exclude fish entirely or apply only generic standards that do not address species-specific physiology (cold tolerance, gill anatomy, neuroanatomy).
Caveats: Regulatory landscape is evolving — the EU and Norway have moved toward stunning requirements for some salmonids, and certifier-driven standards (RSPCA Assured, Global G.A.P.) cover additional throughput voluntarily. The headline figures reflect baseline statutory law, not industry practice plus voluntary schemes.
Schuck-Paim et al. (2025) modeled the cost-effectiveness of switching from air asphyxiation to electrical stunning for rainbow trout slaughter, integrating their Welfare Footprint pain estimates with capital-expenditure data for commercial stunning systems. The result: 60 to 1,200 minutes of moderate-to-extreme pain averted per US dollar of capital expenditure, depending on system, throughput, and farm context. Properly parameterized electrical stunning produces insensibility within 1 to 2 seconds; correctly delivered percussive (priest) stunning produces insensibility in under 1 second. Both are widely recommended by fish-welfare specialists and adopted (variably) in EU farmed salmonid production.
Caveats: Cost-effectiveness depends on system specification, throughput, and the method displaced. Incorrect electrical-stunning parameters (current, frequency, exposure time) can stun without rendering insensible or cause pre-stun stress; equipment must be species-calibrated. Percussive stunning depends on accurate operator delivery.
Sources:
- Schuck-Paim C, Alonso WJ, Pereira PA, Saraiva JL, Cerqueira M, Chiang C, Sneddon LU. Quantifying the welfare impact of air asphyxia in rainbow trout slaughter for policy and practice (2025)
- Sneddon LU, Lopez-Luna J, Wolfenden DCC, Leach MC, Valentim AM, Steenbergen PJ, Bardine N, Currie AD, Broom DM, Brown C. Ample evidence for fish sentience and pain (2018)
The Bottom Line
The quantitative evidence is now clear: typical commercial fish slaughter methods cause minutes of moderate-to-extreme pain per fish, the global scale runs to the trillions per year, and the legal protection gap is near-total. The 2025 Schuck-Paim study moved this question from 'plausible welfare concern' to 'measured welfare disaster'. By any reasonable accounting of suffering, fish slaughter is one of the largest animal-welfare issues on the planet — and electrical or percussive stunning is among the highest-leverage interventions available.
Practical Takeaways
Concrete actions for individuals and supporters: (1) The single largest personal lever is to reduce or eliminate fish in your diet — every fish you don't buy is a fish that doesn't go through this process. (2) If you do eat fish, prefer species and supply chains that use electrical or percussive stunning at slaughter; this is most common for EU-farmed Atlantic salmon and some EU rainbow trout, often signaled by RSPCA Assured, Global G.A.P., or specific producer welfare claims. Verify rather than assume — most farmed fish globally are still asphyxiated. (3) Be skeptical of vague 'sustainably caught' or 'responsibly sourced' labels that say nothing about slaughter method. (4) Support organizations working on humane fish slaughter R&D and policy, including the Welfare Footprint Project and the Fish Welfare Initiative. (5) When advocating, lead with the quantitative numbers — '~10 minutes of pain per fish, 124 billion fish per year, less than 1 percent legally protected at slaughter' is more persuasive than abstract appeals.
Sources & Evidence
5 sources cited across 5 claims
Schuck-Paim 2025: ~10 min median moderate-to-extreme pain per air-asphyxiated trout (95% CrI 1.9-21.7 min)
ModelingTime to LOC in fish slaughter is 2-25 min by species/size/temp; ice slurry can extend it
Modeling~124 billion farmed finfish + 0.78-2.3 trillion wild-caught fish killed for food annually
Observational<1% of farmed fish (by count) have species-specific welfare law; ~70% of production volume has none
ObservationalElectrical/percussive stunning: 1-2 sec to insensibility; 60-1,200 min pain averted per USD capex
Modeling