Most numbers come from published studies. The few that are our own estimates are labelled as such.
How the figures were built
Boundary and unit
Every food figure is kg CO₂e (GWP100) per kg of food as sold, from farm to retail. It includes land-use change, farming, processing, packaging, transport and retail.
It excludes cooking and household food waste.
Figures are global averages, not specific to any country or farm.
Where the food numbers come from
Counting foods…
Judgement calls
Beef mince is the average of the beef-herd and dairy-herd figures, because UK mince mixes both.
Bacon, ham, pork sausages and salami are Poore & Nemecek pork scaled by the processed-to-fresh ratio in the Danish database, to keep one consistent source.
Turkey, duck and chicken thigh use the Poore & Nemecek poultry figure, which doesn't split poultry types.
Dairy derivatives (yoghurt, cream, butter, ice cream, soft and feta-style cheeses) are scaled from Poore & Nemecek milk and cheese using rough multipliers. We didn't use the Danish dairy values because they look inconsistent with the other sources.
Milk chocolate is dark chocolate scaled by cocoa content (30/70).
Tempeh uses dried soybeans as a proxy. Seitan uses wheat × 1.5 for gluten yield.
Brewed coffee and tea are worked back from beans and leaves: 10 g of coffee and 2 g of tea per 250 g brewed.
Where sources disagree, we use Poore & Nemecek when it has the food. Farmed salmon is 13.63 kg CO₂e per kg there, against 1.99 in CONCITO. Wild fish comes from CONCITO, so it looks low by comparison.
Composite dishes
A dish figure is the sum of (ingredient grams × ingredient kg CO₂e per kg) ÷ total ingredient grams. Weights are raw, as bought. Cooking energy and waste are excluded.
Porridge with milk, or with oat milk: oats 40 g, milk or oat milk 200 g, sugar 5 g
Falafel and hummus wrap: tortilla 60 g, falafel 80 g, hummus 30 g, lettuce 20 g, tomato 30 g
Beans on toast: canned beans 130 g, bread 80 g, margarine 5 g
Diet averages and their boundaries
Scarborough et al. (2023) uses LCA data from 570 studies. Its boundary runs from production to retail, including land-use change. It excludes cooking and household waste. Its LCA data differ from Poore & Nemecek, so a plate summed from our foods won't exactly reproduce these diet numbers.
UK average is a proxy: we use the medium-meat diet group. We didn't find an official kg-per-day figure in a primary source.
Global average: 26% of 52.3 Gt CO₂e (about 13.6 Gt) ÷ about 6.96 billion people (2010) ÷ 365 days ≈ 5.35 kg per person per day. It's production-based and includes non-edible by-products and waste.
US average is from Heller et al. (2018): production emissions only, using US-specific LCA factors. It likely understates compared with the UK figure.
EAT-Lancet planetary health diet (3.27 kg per day) is our estimate. The Commission sets intake targets but no emissions figure, so we multiplied its reference intakes by our factors. The intake figures used here haven't yet been checked against the paper.
Equivalents
Car: the UK government factor for an average petrol car, tailpipe only: 0.1645 kg CO₂e per km, so 1 kg ≈ 6.08 km.
Phone: the US EPA figure for one smartphone charge: 0.0124 kg CO₂e, so 1 kg ≈ 80.6 charges.
Known gaps
We found no usable air-freight factor, so air-freighted produce isn't separated.
Palm and soybean oil aren't included, for the same reason.
Tree nuts: Poore & Nemecek's "nuts" figure counts tree carbon uptake, and the CONCITO figures for individual nuts don't, so they differ widely.
Dry foods are per dry kg, and their portion labels say so. Plant milk values are per litre, treated as per kg.
Poore, J. & Nemecek, T. · 2018 · Science 360(6392):987–992
Meta-analysis of 38,700 farms and 1,600 products. Global means per kg of product, farm to retail, read from Our World in Data's CSV (year label 2010). Plant milks are per litre, treated as per kg.
Sheet "GB" (UK retail scenario), column "Total kg CO2-eq/kg". 540 foods, cradle to retail, including indirect land-use change. Our fallback for foods not in Poore & Nemecek. Its dairy values look low next to Poore & Nemecek, so we derive dairy items from Poore & Nemecek instead.
Air-freighted green beans from Kenya/Uganda: about 11 kg CO₂e per kg. Secondary citation: we haven't yet read the original paper, so treat the figure as indicative. The linked report is a related Defra document.
Scarborough, P., Clark, M., Cobiac, L., et al. · 2023 · Nature Food 4, 565–574
EPIC-Oxford, 55,504 adults, diets standardised to 2,000 kcal a day, GWP100, median of the Monte Carlo results (Table 3). Covers production to retail; excludes cooking and household waste.
Ritchie, H. (Our World in Data), based on Poore & Nemecek 2018 · 2019
Food is about 26% of 52.3 Gt CO₂e global emissions (about 13.6 Gt). Livestock and fisheries 31%, crops 27%, land use 24%, supply chain 18%. Source for the global average diet.
Heller, M.C., Willits-Smith, A., Meyer, R., Keoleian, G.A., Rose, D. · 2018 · Environmental Research Letters 13:044004
NHANES 2005–2010 one-day recall. Average US diet production emissions: 4.7 kg CO₂e per person per day (95% CI 4.6–4.8). US-specific LCA factors, so not directly comparable with UK figures.
The Commission gives daily intake targets but no per-person emissions figure. The 3.27 kg CO₂e per day planetary health figure is our estimate: its reference intakes multiplied by our factors. Based on a 2,500 kcal reference diet. The intakes haven't yet been checked against the paper.
Factors we derived from the sources above using documented scaling rules, such as brewed coffee from beans, dairy derivatives from Poore & Nemecek milk and cheese, and processed pork from Poore & Nemecek pork. Treat as estimates.
Each dish is the weighted average of its ingredients' factors. These are illustrative single recipes, not measured products. Cooking energy is excluded.
Every food
Search or filter all the foods behind the site. Click a column heading to sort it, or a food name to add it to your plate.