Food cost percentage is the share of your food sales that went on the food itself. The formula fits on one line:
(beginning inventory + purchases − ending inventory) ÷ food sales × 100
Everything in that line is easy to find except the two inventory counts. That's where most restaurants' number goes wrong, and it goes wrong quietly, because a bad count still produces a believable percentage.
The formula, worked
Take a month. You started with $8,400 of food on the shelves, bought $31,000 more, and ended with $7,900 left.
- Food used: $8,400 + $31,000 − $7,900 = $31,500
- Food sales for the month: $102,000
- Food cost percentage: $31,500 ÷ $102,000 × 100 = 30.9%
The middle step is the whole point of the formula. What matters is the food you used, not the food you bought. Divide purchases alone by sales and a big delivery on the 30th makes this month look terrible and next month look brilliant, when nothing about the kitchen changed.
What is a good food cost percentage?
Most published guides put a restaurant somewhere between 28% and 35%. The National Restaurant Association's survey of operators gives a firmer figure: a median of 32% of sales for full-service restaurants in 2024, counting non-alcoholic drinks with the food, and 32.4% for limited service. Quick service, in other words, doesn't automatically run cheaper on food. Volume mattered more. Full-service restaurants with more than $2 million in sales came in at 31%, smaller ones at nearly 34%.
Treat any of these as a way to notice that something has moved, not as a target. The band is wide because menus are: a bowl of pasta and a plate of fish don't cost the same share of their price, and neither kitchen is doing anything wrong.
Two rules make the number comparable at all:
- Keep food and beverage apart. Drinks, spirits and beer especially, usually cost a smaller share of their price than food does, so a blended figure improves every time the bar has a good night, and hides a kitchen problem in the process.
- Use the same sales figure your profit and loss uses, net of discounts and refunds. The same rule applies to labor cost, for the same reason.
Where the number goes wrong
The count. Counting on a different day each month, or at a different time of day, is enough to move the result. So is mixing units — a case in one count and single items in the next — or valuing stock at last quarter's prices after a supplier has raised them. The size of the error is easy to underestimate: on $102,000 of sales, a count that's out by $1,000 moves your food cost by about a full point.
Purchases in the wrong month. Invoices dated on the 29th for food that arrived on the 2nd. Credits for short deliveries and returns that never get taken off. Each one shifts cost from one month into another.
Food that left the shelf without being sold. Comps, spoilage, waste and the plate that went back all use food and earn nothing, and all of them belong in food cost. Staff meals are the exception most cost-control textbooks make: they come out of food cost and are usually counted with labor, as a staff benefit. Either treatment is defensible, as long as you pick one and keep to it.
The fix for the first two is boring and it works: count on the same day, at the same time, in the same order, with the same units, and book every invoice to the month the food arrived.
Actual versus theoretical food cost
The formula above gives you your actual food cost. There's a second number worth having beside it.
Theoretical food cost is what the month should have cost, given what you sold. Multiply each dish's plate cost by how many you sold, add it up, and divide by food sales. It's the figure you'd get if every portion were exact and nothing was wasted, comped or taken home.
The gap between the two is the useful part. Restaurant software companies describe a gap of about a point as best in class. The standard cost-control textbook is more forgiving: it treats an actual cost of up to 110% of the theoretical one — 33% against a theoretical 30% — as achievable, and anything beyond that as a sign of waste, theft, spoilage, over-portioning, or recipe costs that have gone stale because a supplier raised prices and nobody updated the plate cost.
You need itemized sales to do this — how many of each dish, not just the total. Without them you have one number and no way to see inside it.
Using it to price a dish
The percentage turns into a pricing rule when you run it backwards:
menu price = plate cost ÷ target food cost percentage
A dish with $4.20 of ingredients, at a 30% target, prices at $4.20 ÷ 0.30 = $14.00.
It's a good starting point and a poor finishing one, because a percentage is not money. A $30 steak at 38% food cost leaves $18.60 after ingredients. A $14 pasta at 22% leaves $10.92. The steak looks worse on the percentage and puts more into the drawer every time it sells. Menu engineering, the approach developed at Michigan State in the early 1980s, ranks dishes on exactly that — the dollars each one contributes and how often it sells — rather than on food cost percentage. Price with the percentage, then check the dollars.
Put it next to labor
Food cost is half of the number that actually governs a restaurant. Prime cost — food plus labor — matters more than either part, because the two trade directly against each other. Make everything in house and your food cost falls while your hours rise; buy prepared components and it reverses.
So a food cost that improved by two points in the same month the kitchen added a prep shift hasn't necessarily improved at all. Your labor cost is 32%. Is that good? covers the other half, and the restaurant labor cost calculator works out labor fully loaded, so the two halves are measured the same way. And if the question underneath all of this is how much is left at the end of the month, restaurant profit margins follows the dollar the rest of the way down.