Plain language summary

Ketosis is the state your body gets into when you restrict your carbs to very low levels (typically eating less than 50 grams per day)
A high protein intake (i.e. 2.2 grams per kilogram of body weight) does not seem to prevent ketosis, as suggested by many studies. In fact, there is a macronutrient which is much more important than protein for ketosis. Namely:
Carbohydrates. A very low carbohydrate diet will put you in a state of ketosis. We’ve yet to find out exactly how many carbs you can eat while staying in ketosis. People have maintained ketosis on 0-80 grams of carbs per day, in many keto studies. Aim for 20-60 grams of carbs if you want to ensure ketosis.

Will ketosis help you burn fat compared to a calorie-matched alternative diet? Read our article here.

What is ketosis?

Ketosis takes place when your body lacks carbohydrates as a primary fuel source (Adam-Perrot et al., 2006; Westman et al., 2007). Your body then adapts by using fat and ketones for fuel (Laffel, 1999; Adam-Perrot et al., 2006; Westman et al., 2007; Paoli et al., 2013; Paoli, 2014). It is a natural response to starvation, extended fasting, and prolonged exercise (Mitchell et al., 1995; Laffel, 1999; VanItallie and Nufert, 2003; Cox and Clarke, 2014; Newman and Verdin, 2017; Puchalska and Crawford, 2017). Yet, ketosis is not the same as ketoacidosis which is a much more severe form of ketosis (Laffel, 1999; Puchalska and Crawford, 2017).

There are three ketone bodies:

β-hydroxybutyrate [BHB]
Acetoacetate [AA]
Acetone

You can measure them in blood, urine, or breath (Laffel, 1999; Clocker et al., 2013; Anderson, 2015; Urbain and Bertz, 2016). A BHB level of 0.3 mmol/l indicates that you are in mild ketosis. BHB measurement is generally more stable and reliable than urinary ketone measurement because the latter fluctuates more (Urbain and Bertz, 2016). I.e. your 2PM urinary measurement might be very different from your 2AM measurement.

Does a high protein diet prevent ketosis?

Several keto websites like Diet Doctor and Perfect Keto suggest that a high protein diet could prevent ketosis. The rationale is that protein can be turned into glucose and this in turn could take you out of ketosis. They ask you to restrict protein to 1-1.6 grams per kilo body weight.

In other words, if you weigh 70 kg, they claim you should eat no more than 70-112 grams of protein per day.

“Some people can get away with eating more protein, but others might need to keep it at 15% or less to prevent gluconeogenesis and stay in ketosis. In terms of grams, this will usually fall around 0.7 to 0.9 grams of protein per pound of lean bodyweight (or 1.5 to 2.0 grams of protein per kilogram of lean bodyweight).” - Perfect Keto

To simplify their calculation, let’s say a man has 20% body fat. His fat-free mass would therefore be 80%. If his body weight is 80 kg, his fat-free mass would be 64 kg. Going by the numbers above, he could not eat more than 2 grams of protein per kg of lean weight (128 grams of protein in this case, which translates into 1.6 grams per kg body weight).

“Restrict protein to moderate levels. If possible stay at or below 1 gram of protein per day, per kg of body weight (0.45 grams per pound). So about 70 grams of protein per day if you weigh 70 kilos (154 pounds).” - Diet Doctor

So what does the scientific literature have to say, is there a protein threshold where eating more protein prevents or limits ketosis?

Protein and ketosis table

Protein and ketosis sparkline chart

Protein and ketosis correlation

Below is a table of keto studies that reported daily protein and carb intakes, and ketone levels. The data is already shown in the other graphs, but some people might prefer to have the data presented this way.

Only high-protein studies are included to keep the table relatively compact (i.e. studies with more than 100 g protein per day).

But if this chart is made of studies, where are they? Look at the first row in the chart below (BHB, mmol/l). Every peak and trough represents one study mean (data from one group). You could draw a straight vertical line from top to bottom, and that line would represent related data from one group of participants. So you can see that the group of participants with 4 mmol/l BHB (the highest blue peak) also had a non-existent carbohydrate intake, and also a low protein intake (about 70 grams).

This chart is sorted by protein which is in the bottom row (less->more). You can use this visualization to check whether there is an apparent link between daily protein intake and measured BHB levels. You will find the same chart in the carb section [internal link], sorted by BHB (low->high).

Every blue dot represents a study mean (a group of participants from a study). The y-axis shows how much protein the group of participants ate on average per day. The x-axis shows BHB levels measured during the study.

Running the protein data through Pearson’s linear correlation gives us R = -0.26, a weak negative linear correlation. r^2 = 0.07. The correlation is fully contingent on the Bogardus et al. (1981) deep keto group, the blue dot to the very right in the graph.

If we remove Bogardus et al. (1981), the direction of the correlation is reversed (R = 0.17, r^2 = 0.03):

How your carb intake affects ketosis

There is consensus in the keto literature that a very low carb diet induces ketosis. Most researchers recommend eating less than 50 grams of carbs to enter ketosis (Adam-Perrot et al., 2006; Westman et al., 2007; Accurso et al., 2008; Bueno et al., 2013; Paoli et al., 2013; Gregory et al., 2017). There are various websites recommending even less carbs. This has caused a lot of confusion as to how many carbs you can actually eat to stay in ketosis.

We have decided to answer this question by analyzing data from the ketogenic literature.

Carbs and ketosis sparkline chart

Carbs and ketosis non-linear model

Zoomed carb model (0-150g)

Linear carb ketosis correlation

Carbs and ketosis table

This chart shows how many carbs and protein people ate in various studies. Their BHB levels were also recorded. The chart below is visualized based on average numbers from keto or control study groups (study means). It is the same chart presented above, except sorted by BHB levels (top row). You can use this visualization to check whether there is an apparent link between daily carb/protein intake and measured BHB levels.

Every blue dot represents a group of participants from a study. The y-axis shows how many grams of carbs the group of participants ate on average per day. The x-axis shows BHB levels measured during the study.

r^2 = ~-0.7 (non-linear)

The data in this scatterplot hints at how important carbohydrates are for ketosis. The relationship seems to be non-linear. A correlation cannot tell us that there is a dose-response relationship by itself, but we already have evidence from studies suggesting a causal link between carbs and ketosis. However, note that our data here is based on averages and not individual data (a limitation).