The short answer
Breathe into the device and it tells you whether you’re “burning fat or sugar”: the promise of breath analyzers is appealing. Depending on the model, they measure acetone or CO₂—two different things—and none can see your fat loss.
A metabolic breath analyzer can be useful for tracking a trend under consistent conditions. It cannot reliably reveal your “true energy source,” your fat loss or your overall metabolic health.
Two types of device are often confused. Acetone sensors track a volatile compound produced during ketogenesis. CO₂-based devices attempt to estimate whether oxidation at that moment leans more toward fats or carbohydrates. They therefore measure different things, and their scores are not interchangeable.
My verdict is simple: the tool can answer a narrow experimental question, such as “does my morning reading change when I adjust my carbohydrate intake?” On its own, it does not justify prolonged fasting, tighter restriction or buying supplements.
Acetone, BHB and CO₂: three different signals
When the liver converts fatty acids into ketone bodies, it produces acetoacetate and beta-hydroxybutyrate, or BHB, among others. Some acetoacetate breaks down into acetone, which passes into exhaled air.
This biological connection does not allow a universal conversion between parts per million of acetone and mmol/L of BHB. The clinical review by Blaikie and colleagues describes important technical and biological limitations: recent food intake, exercise, ventilation, medication, individual variability and sampling standardization. A breath reading should therefore not be converted into “depth of ketosis” using a fixed rule.
CO₂ works differently. In indirect calorimetry, respiratory quotient is the ratio of CO₂ produced to oxygen consumed. At steady state, a lower value is consistent with a greater contribution from fats, while a higher value suggests more carbohydrates. But a device that does not simultaneously measure oxygen does not perform the same measurement as a laboratory metabolic cart. It produces a model-based estimate.
The practical distinction is this:
- an acetone sensor mainly provides information about ketogenesis
- a CO₂ device estimates the substrate mix at the time of testing
- neither directly measures the body fat used over a day
- no score diagnoses insulin resistance or ketoacidosis
What the human evidence actually shows
Exhaled acetone is a plausible, noninvasive biomarker. A 2015 review observed that it increased with carbohydrate restriction, fasting and weight loss in several experiments. It also notes that meal composition, exercise and breathing affect the result. This makes it a context-dependent signal, not a portable energy-balance meter.
The 2021 review of ketone body measurement reaches a similar conclusion: blood BHB and exhaled acetone each have value, but ranges vary substantially between nutritional ketosis, ketogenic therapy and ketoacidosis. Applicability depends on the clinical question.
For consumer CO₂ analyzers, the published evidence remains much thinner than the marketing. Internal validation or average agreement with a laboratory device does not demonstrate that a daily score improves weight, HbA1c, performance or cardiovascular outcomes. Three levels of evidence need to be distinguished: signal validity, algorithm accuracy and the benefit of a decision guided by that algorithm. The third level is largely missing.
What the score cannot tell you
“I’m burning fat, so I’m losing weight”
Your body oxidizes fats throughout the day, even without losing body fat. After a low-carbohydrate meal, the proportion of fat used at that moment may increase while a calorie surplus prevents any net loss. Conversely, someone can lose fat over several weeks without showing marked ketosis.
The most useful variables for assessing fat loss remain:
- seven-day average weight
- waist circumference measured consistently
- intake and activity tracked over several weeks
- changes in performance, hunger and sleep
“My low score proves my metabolic flexibility”
Metabolic flexibility means the ability to adjust substrate use according to fasting, meals and activity. A single measurement on waking is not a validated test of it. A reading may also reflect dinner, alcohol, the previous day’s exercise, ventilation during the test or poor technique.
“I can rule out ketoacidosis”
This is the most dangerous conclusion. In someone with diabetes, especially when taking an SGLT2 inhibitor, ketoacidosis can occur with lower blood glucose than expected. Nausea, vomiting, abdominal pain, rapid breathing, confusion or feeling unwell require urgent assessment. A consumer device replaces neither blood BHB testing nor medical advice.
How to compare devices without buying into the marketing
Before buying, check observable criteria:
- the analyte actually measured: acetone or CO₂
- availability of raw results, rather than only an opaque score
- a validation study on the exact commercial model
- a relevant comparator: spectrometry or indirect calorimetry, depending on the analyte
- repeatability across several consecutive breaths
- calibration procedure and sensor lifespan
- operation without a subscription and the ability to export data
- privacy policy and what happens to the device if the service shuts down
One brand’s color thresholds should not be applied to another. Semiconductor sensors may also react to alcohol or other volatile compounds. A promise of accuracy without reporting error, limits of agreement and the population tested is not validation.
Fourteen-day testing protocol
The mechanism requires controlling food intake, time since the last meal, exercise and ventilation. Otherwise, you’re mostly measuring noise.
Use this cautious protocol:
- measure on waking, after using the bathroom, before coffee, calories, brushing your teeth or exercise
- follow the manufacturer’s breathing maneuver exactly, without hyperventilating
- take two measurements two minutes apart during the first three days to assess repeatability
- keep your diet and dinner time fairly consistent during the first week
- change only one variable during the second week, such as the amount of carbohydrate at dinner
- record sleep, alcohol, training and any illness that could explain an unusual reading
- compare weekly medians, not the best isolated score
Stop the experiment if it pushes you to prolong a fast despite feeling unwell, dizziness or other symptoms. Reassess after fourteen days. If no reproducible relationship emerges with your chosen variable, the device probably does not provide useful information for your decisions.
To understand ketosis beyond a score, the guide to the targeted ketogenic diet and performance · FR provides a more useful nutritional framework.
The Verdict
An acetone analyzer can reduce the cost and burden of repeated ketogenesis monitoring. A CO₂ device can provide behavioral biofeedback. Their usefulness is real but narrow.
I wouldn’t buy this tool to “optimize longevity” or prove fat loss. I would consider it only if the question were defined before purchase, the model had an identifiable published validation and the decision could be confirmed using more robust measurements. Proper use is not about chasing the lowest score. It is about checking whether a reproducible trend helps you make a decision without increasing risk.
