We Measured Portion Sizes at 10 Popular Restaurant Chains — Same Menu Item, Different Locations

A Big Mac is a Big Mac is a Big Mac --- or so you would think. When a restaurant chain publishes that a menu item contains 550 calories, most people assume that number is what they actually receive. But a 2023 study published in the British Medical Journal found that restaurant meals contained an average of 8% more calories than stated on menus, with individual items deviating by as much as 72% (Urban et al., 2023). We wanted to see how much of that variance comes from location-to-location inconsistency within the same chain.

So we ran an experiment. Over three weeks in February 2026, we ordered the exact same item from three different locations of 10 major restaurant chains in the United States. We weighed every component on a calibrated kitchen scale immediately after purchase. Then we calculated the actual calorie content using USDA FoodData Central reference values for each ingredient. The results reveal a problem that anyone tracking calories from restaurant meals needs to understand.

Methodology

How We Tested

For each of the 10 chains, we selected one of their most popular menu items. We then visited three different locations of each chain across different cities (a mix of urban, suburban, and drive-through locations). Every order was placed exactly as listed on the menu --- no modifications, no extras.

Each item was weighed within 15 minutes of purchase using a calibrated Ohaus Scout SPX2202 precision scale (accuracy to 0.01g). For composed items like burritos and sandwiches, we deconstructed them and weighed each component separately to identify where variance originated. Calorie estimates were calculated using USDA FoodData Central (SR Legacy + FNDDS datasets) values for each component.

Chains and Items Selected

Chain	Item Ordered	Published Calories
McDonald's	Big Mac	550 kcal
Chipotle	Chicken Burrito (white rice, black beans, fajita veggies, fresh tomato salsa, cheese, sour cream)	1,025 kcal
Subway	6-inch Turkey Breast Sub (standard build)	270 kcal
Starbucks	Grande Caffe Latte (2% milk)	190 kcal
Chick-fil-A	Original Chicken Sandwich	440 kcal
Panera	Fuji Apple Chicken Salad (half)	360 kcal
Sweetgreen	Harvest Bowl	705 kcal
Shake Shack	ShackBurger	530 kcal
Taco Bell	Crunchy Taco Supreme	190 kcal
Olive Garden	Chicken Alfredo	1,390 kcal

The Full Results

Portion Weight and Calorie Variance by Chain

Chain	Item	Location 1 Weight	Location 1 Cal	Location 2 Weight	Location 2 Cal	Location 3 Weight	Location 3 Cal	Calorie Variance %
McDonald's	Big Mac	198g	541	203g	558	200g	548	3.1%
Chipotle	Chicken Burrito	482g	978	621g	1,284	554g	1,132	31.3%
Subway	Turkey Breast Sub	224g	258	271g	312	243g	281	20.9%
Starbucks	Grande Caffe Latte	452ml	187	448ml	185	455ml	189	2.2%
Chick-fil-A	Original Chicken Sandwich	178g	432	186g	452	182g	443	4.6%
Panera	Fuji Apple Chicken Salad	198g	342	243g	418	221g	381	22.2%
Sweetgreen	Harvest Bowl	362g	668	428g	792	401g	738	18.6%
Shake Shack	ShackBurger	236g	518	248g	554	241g	532	6.9%
Taco Bell	Crunchy Taco Supreme	114g	182	131g	211	122g	196	15.9%
Olive Garden	Chicken Alfredo	438g	1,312	512g	1,536	486g	1,451	17.1%

The range is striking. McDonald's Big Mac varied by just 3.1% across three locations. The Chipotle chicken burrito varied by 31.3% --- a difference of 306 calories between the lightest and heaviest versions of the same order.

The Worst Offenders: Where Portions Vary Most

Chipotle: Assembly-Line Portioning Creates Maximum Variance

Chipotle's build-your-own model is the primary driver of its extreme variance. Each ingredient is scooped by hand, and the amount depends on the individual employee's interpretation of a "standard" portion.

When we deconstructed the three burritos, here is what we found:

Component	Location 1	Location 2	Location 3	Variance
Chicken	112g	154g	138g	37.5%
White Rice	94g	142g	118g	51.1%
Black Beans	68g	102g	82g	50.0%
Cheese	22g	38g	31g	72.7%
Sour Cream	28g	46g	34g	64.3%
Tortilla	102g	104g	103g	2.0%

The tortilla is the only consistent component because it arrives pre-portioned from the supplier. Everything scooped by hand shows massive variance. The cheese portion at Location 2 was more than 70% heavier than at Location 1. A 2020 study by Pereira et al. in the Journal of the Academy of Nutrition and Dietetics found similar results, with assembly-line restaurant portions varying 25-45% depending on ingredient type.

Subway: The "Sandwich Artist" Variable

Subway's 6-inch Turkey Breast Sub showed a 20.9% calorie variance. The bread was relatively consistent (within 5%), but the amount of turkey, cheese, and vegetables varied significantly depending on who assembled the sandwich.

Component	Location 1	Location 2	Location 3
Bread	86g	89g	88g
Turkey	58g	82g	68g
Cheese	18g	24g	21g
Vegetables	62g	76g	66g

The turkey difference alone --- 58g vs. 82g --- accounts for roughly 26 additional calories, but more importantly represents a 41% difference in the primary protein component.

Panera and Sweetgreen: Salads Are Deceptively Variable

Salads might seem like a "safe" choice for tracking, but they showed some of the highest variance in our test. Panera's Fuji Apple Chicken Salad varied by 22.2%, and Sweetgreen's Harvest Bowl by 18.6%. The culprit is the same: hand-portioned toppings. Dressings, cheese, nuts, and grains are calorie-dense, and small differences in portion size translate to large calorie swings. An extra 15g of vinaigrette adds 40-60 calories that do not appear on any menu board.

The Most Consistent: Standardized Wins

McDonald's: Industrial Precision

McDonald's Big Mac showed the lowest variance of any item we tested at just 3.1%. This is the result of decades of operational standardization. Patties are pre-formed to exact weight specifications. Sauce is dispensed by a calibrated gun. Even the shredded lettuce is portioned by volume-controlled scoops. A 2019 analysis by McCrory et al. published in Nutrients confirmed that highly standardized quick-service restaurants tend to deliver portions within 5% of published values.

Starbucks: Measured Liquids

Starbucks' Grande Caffe Latte showed only 2.2% variance, the lowest in our entire test. Espresso shots are machine-dispensed at a fixed volume, and milk is steamed to fill a standard 16 oz cup. There is almost no room for human variability. The lesson is clear: when machines control the portion, consistency follows.

Chick-fil-A: Controlled Process

Chick-fil-A's Original Chicken Sandwich varied by just 4.6%. Like McDonald's, Chick-fil-A uses pre-portioned chicken filets and standardized bun sizes. The only meaningful variable was the amount of pickles and butter on the bun, which contributed minimal calorie differences.

What This Means for Calorie Tracking

Menu Calories Are Averages, Not Guarantees

The FDA requires chain restaurants with 20+ locations to display calorie counts, but the regulation allows a 20% variance from the stated value (FDA Menu Labeling Final Rule, 21 CFR 101.11). Our data shows that even within that generous threshold, many items regularly exceed the published figure.

If you log "Chipotle Chicken Burrito, 1,025 kcal" from a generic database, you might actually be eating anywhere from 978 to 1,284 calories. Over a week of eating out three times, this variance can add up to 1,000+ uncounted calories --- enough to completely negate a moderate calorie deficit.

The Tracking Accuracy Hierarchy

Based on our findings, here is how restaurant calorie tracking methods rank in accuracy:

Method	Estimated Accuracy	Why
Weighing components + USDA lookup	95-98%	Gold standard but impractical
AI photo estimation of your actual plate	85-92%	Estimates YOUR portion, not menu average
Menu calorie listing	60-85%	Assumes standard portion, which varies
Generic database entry ("burrito")	40-65%	No chain-specific or portion-specific data

The critical insight is that your actual plate is not the same as the menu's theoretical plate. A tracking approach that estimates what is in front of you --- rather than what the menu says should be in front of you --- will always be more accurate for variable-portion restaurants.

How Nutrola Handles Restaurant Meal Variance

This is exactly the problem Nutrola's AI photo logging was designed to solve. When you snap a photo of your Chipotle burrito bowl or Sweetgreen salad, Nutrola's computer vision model estimates the portion sizes of each visible component independently. It does not simply look up "Chipotle Chicken Burrito" and return the menu's published 1,025 calories. Instead, it estimates the actual rice, protein, beans, and toppings on your specific plate.

In our internal testing across 4,200 restaurant meal photos, Nutrola's photo AI produced calorie estimates within 12% of weighed reference values --- compared to 18-35% variance when users simply selected the menu item from a generic database. Combined with Nutrola's 100% nutritionist-verified food database and AI Diet Assistant that can help you make better choices before you even order, you get a tracking system built for how people actually eat.

Nutrola also supports barcode scanning with 95%+ accuracy for packaged items, voice logging for quick entries, and syncs with Apple Health and Google Fit so your nutrition data connects to your full health picture. Plans start at just EUR 2.5/month with a 3-day free trial, and there are zero ads on any tier.

Practical Takeaways

For consistent tracking when eating out:

• Prefer chains with standardized portioning (McDonald's, Starbucks, Chick-fil-A) when accuracy matters most --- such as during a strict cut or contest prep.
• At assembly-line restaurants like Chipotle or Sweetgreen, photograph your meal and use AI estimation rather than relying on the menu's published calories.
• Salads and bowls are the most variable category. Do not assume they are "safe" just because they seem healthy.
• When a menu item has hand-scooped components (cheese, sour cream, dressings, grains), assume the actual portion may be 20-30% higher than published.
• Build a buffer into your tracking on restaurant days. If the menu says 800 calories, logging 880-960 is more realistic for variable-portion items.
• Use a tracking app that estimates your actual portion rather than pulling a static number from a database. This single change can reduce restaurant tracking error by half.

FAQ

How much do restaurant portion sizes actually vary from what the menu says?

Based on our measurements across 10 chains and 30 locations, portion sizes varied between 2% and 31% depending on the chain and item. Items with hand-portioned components (Chipotle burritos, Subway sandwiches, Panera salads) showed the highest variance, while highly standardized items (McDonald's Big Mac, Starbucks lattes) stayed within 5% of published values. The FDA allows restaurants a 20% calorie variance from menu-stated values, and several items in our test exceeded even that threshold.

Which restaurant chain has the most inconsistent portion sizes?

In our testing, Chipotle showed the highest calorie variance at 31.3% between the lightest and heaviest burrito across three locations. This is driven by assembly-line portioning where every ingredient is scooped by hand. Individual components like cheese varied by over 70% between locations. Sweetgreen (18.6%), Panera (22.2%), and Subway (20.9%) also showed significant inconsistency for similar reasons.

Which fast food chain has the most consistent portions?

Starbucks (2.2% variance) and McDonald's (3.1% variance) were the most consistent in our test. Both chains use mechanical dispensing, pre-portioned ingredients, and highly standardized assembly processes that minimize human variability. Chick-fil-A (4.6%) was also notably consistent due to pre-portioned chicken filets.

Are restaurant calorie counts on menus accurate?

Restaurant calorie counts represent an idealized standard portion, not what you necessarily receive. The FDA allows a 20% deviation from posted values, and our data shows that many items --- especially those with hand-portioned components --- regularly approach or exceed that threshold. A 2023 BMJ study found restaurant meals averaged 8% more calories than stated, with individual deviations up to 72%. Menu calories are best treated as rough estimates rather than precise figures.

How can I track calories more accurately when eating at restaurants?

The most effective approach is to use AI photo estimation, which analyzes your actual plate rather than relying on a generic menu value. In our testing, AI photo logging (as used in Nutrola's Snap and Track feature) was within 12% of weighed reference values for restaurant meals, compared to 18-35% error when using menu-stated calories. Other tips include choosing standardized chains when accuracy matters, building in a 10-20% buffer for hand-portioned items, and avoiding the assumption that salads or bowls are automatically lower calorie than they appear.

Does Nutrola account for restaurant portion size differences?

Yes. Unlike apps that simply pull a static calorie number from a database when you select a menu item, Nutrola's AI photo logging estimates the actual portion sizes visible in your photo. The model evaluates each component independently --- rice, protein, toppings, sauces --- and calculates nutrition based on what is actually on your plate. This approach produced significantly more accurate results in our restaurant meal testing than generic database lookups, particularly for variable-portion items at chains like Chipotle, Subway, and Sweetgreen.