Early-morning hormones make the liver release more glucose while insulin works less well. Here is what the research shows, what else can raise a waking reading, and how to collect data your care team can use.
You go to bed at 6.5 mmol/L (117 mg/dL), eat nothing overnight, and wake up at 9.5 mmol/L (171 mg/dL). On a continuous glucose monitor (CGM), the line often sits fairly flat until the early hours and then climbs steadily until you get up. That shape has a name. The dawn phenomenon is an early-morning rise in glucose, or in the insulin needed to keep glucose steady, that happens without a low beforehand 1.
It is common in both type 1 and type 2 diabetes 1, and in type 2 it adds measurably to average glucose and A1c 2.
A high waking number can also come from dinner, a bedtime snack, background insulin that fades before morning, or a night-time low treated with extra carbohydrate. This article covers what drives the dawn rise, how large it tends to be, what the evidence says about the Somogyi effect, which approaches care teams consider, and how to collect data that help tell these patterns apart.
The name appeared in the title of a 1981 study of 11 people with type 1 diabetes who were monitored over 24 hours 3. Ten of the 11 had an early-morning rise in fasting glucose, and the size of that rise tracked with how high glucose went after breakfast 3.
A few years later, researchers defined it as an abrupt increase in fasting glucose, insulin needs or both between about 5 and 9 a.m., with no low beforehand 1. Studies that use CGM now usually measure it as the change from the lowest overnight value (the nadir) to the reading just before breakfast. A rise of about 1.1 mmol/L (20 mg/dL) or more is a common research threshold for saying it is present 2, 4.
Two details matter for your own data. A rise that follows a night-time low is a separate pattern, covered below. And the rise varies from night to night. In five people with type 1 diabetes who were each studied on four nights, it appeared on 17 of the 20 nights, and its size changed between nights 1.
The early-morning rise starts as normal human physiology. In a 1984 study, researchers measured glucose, insulin and hormones every half hour between 1 a.m. and 9 a.m. in eight volunteers without diabetes 5. After about 5:30 a.m., the liver released more glucose, and glucose, insulin and C-peptide (a marker of the body’s own insulin release) all rose as the pancreas secreted more insulin to match 5. Cortisol, adrenaline and noradrenaline climbed between 4:00 and 6:30 a.m., after pulses of growth hormone earlier in the night 5. The authors concluded that early-morning increases in glucose and insulin needs in diabetes may be an exaggeration of a normal daily rhythm in how strongly the liver responds to insulin 5.
The clearest test came from a 1985 study of six people with type 1 diabetes who received a constant overnight insulin infusion, so their insulin supply could not change 6. Glucose rose from 5.4 mmol/L (98 mg/dL) at midnight to 12.5 mmol/L (225 mg/dL) at 8 a.m. Glucose production rose by 65%, and the rate at which glucose was cleared from the blood fell by 50% 6. When the researchers suppressed the night-time growth hormone surges with a somatostatin infusion (replacing the glucagon it also suppresses), glucose and glucose production no longer rose. When they added growth hormone back in hourly injections, the rise returned. Blocking the effects of adrenaline and noradrenaline made no significant difference 6.
A 2013 review of 30 years of this research, from the University of Perugia in Italy, named night-time growth hormone spikes as the most likely mechanism in type 1 diabetes 7. Cortisol also rises toward morning 5, but in the key experiments growth hormone was the hormone whose removal stopped the rise 6.
Put together, toward dawn the liver releases more glucose while the body’s tissues take up less of it, for the same amount of insulin 6.
A body without diabetes answers the extra early-morning glucose with extra insulin 5. In type 1 diabetes, and in type 2 when the body’s own insulin response is reduced, that automatic answer is weak or missing, so injected or pumped insulin has to cover the extra need.
Timing adds a second problem. In the 1981 study, free insulin was high relative to glucose during the night, when three of the four recorded lows happened, and lowest relative to glucose at 8 a.m. 3. Insulin needs shift through the night, from a low point in the early hours to a peak around dawn 8, so a fixed overnight dose that fits 7 a.m. can be more than the body needs at 2 a.m. NPH, an older intermediate-acting insulin, adds to the mismatch when taken in the evening, because it peaks early and then wanes 7.
In plain terms: Toward morning, hormones tell your liver to release more glucose and make insulin work less well. A pancreas that works normally answers with more insulin. When your insulin comes from a pen or a pump, the dose has to cover that extra need, and if it falls short or fades before morning, glucose climbs.
The largest early study measured overnight insulin needs in 114 people with type 1 diabetes, adjusting an intravenous insulin infusion through the night to keep glucose normal 8. Insulin needs fell after midnight to a low point around 2:40 a.m. and then climbed to a peak around 6:40 a.m. An increase of more than 20% lasting at least 90 minutes, the study’s definition of the dawn phenomenon, was present in 101 of the 114 people (89%) 8. On average, insulin needs between 5 and 7 a.m. were about 19% higher than between 1 and 3 a.m. The pattern repeated on three separate nights, and it was smaller in people who had lived with diabetes for longer 8.
In the 1984 study that used feedback-controlled insulin infusions, insulin needs rose by at least 50% for an hour and a half in the early morning in 75% of people with type 1 diabetes and 77% of people with type 2 diabetes 1.
CGM made it possible to measure the rise in everyday life. In a study of 248 people with type 2 diabetes who were not treated with insulin and wore CGM, the median rise from the overnight low to the pre-breakfast reading was 0.9 mmol/L (16 mg/dL) 2. The spread was wide. A quarter of participants had no measurable rise, and a quarter rose by 1.75 mmol/L (31.5 mg/dL) or more 2.
The researchers estimated that the dawn rise added 0.39% (4.3 mmol/mol) to A1c on average, and about 0.7 mmol/L (12.4 mg/dL) to 24-hour mean glucose 2. Its effect was similar in people managing with diet alone and in those taking oral medications, and the authors concluded that none of the oral medications available at the time eliminated it 2.
The 2013 review described the dawn phenomenon as a night-to-morning rise in glucose before breakfast and, to a larger extent, after it 7. A 2026 CGM study from Huadong Hospital, Fudan University, in Shanghai looked at what its authors call the extended dawn phenomenon in 460 adults with type 2 diabetes and an A1c below 7.5% 4. People with a dawn rise of at least 1.1 mmol/L (20 mg/dL) had higher peaks after breakfast than people without one, a median of 9.7 versus 8.9 mmol/L (175 versus 160 mg/dL), even though their fasting glucose was similar. The size of the dawn rise predicted high post-breakfast glucose independently of A1c and measures of insulin resistance 4. Because it looked at one point in time, the study shows an association and cannot prove cause.
Average overnight insulin needs in 114 people with type 1 diabetes. The two marked points are published averages; the curve between them is simplified.
The Somogyi effect is the idea that a low during the night triggers a surge of insulin-opposing hormones, which then drives glucose high by morning 9. The dawn phenomenon and the Somogyi effect point treatment in opposite directions, which is why researchers have tested the idea carefully.
Several studies found little support for it. In a 1980 study, 15 people treated with insulin had 17 untreated, symptom-free lows below 2.0 mmol/L (36 mg/dL) between 11 p.m. and 3 a.m. 9. The six people who woke above 7 mmol/L (126 mg/dL) did not have higher growth hormone, cortisol or glucagon than the others. Their morning glucose tracked closely with their insulin levels instead, which pointed to insulin running low late in the night 9.
In a 1987 study of 10 people with type 1 diabetes on their usual treatment, glucose was deliberately brought down to about 2.0 mmol/L (36 mg/dL) between 2 and 4:30 a.m. on one of three nights 10. The 8 a.m. reading averaged 6.3 mmol/L (113 mg/dL), no higher than on a night when lows were prevented, 10.1 mmol/L (182 mg/dL), or on an ordinary night, 8.3 mmol/L (149 mg/dL). On the ordinary night, a lower overnight low point went with a lower morning reading 10.
A 1988 study found a different result. In 10 people with type 1 diabetes using insulin pumps, extra insulin given between 10 p.m. and 2 a.m. to mimic an overdose produced symptom-free lows of about 2.4 mmol/L (43 mg/dL) 11. Compared with nights when glucose was kept from falling, fasting glucose was higher, 7.3 versus 6.2 mmol/L (131 versus 112 mg/dL), and so was glucose after breakfast, 12.5 versus 8.7 mmol/L (225 versus 157 mg/dL). The authors concluded that in people on intensive insulin therapy, a night-time low can worsen next-morning control and should be considered when morning highs are unexplained 11.
CGM studies have since looked at thousands of nights. In records from 2,600 adults with type 2 diabetes on stable insulin treatment, mornings after a night-time low (below 3.9 mmol/L, or 70 mg/dL, for at least 15 minutes between midnight and 6 a.m.) had lower fasting glucose than mornings without one, across 4,705 nights of usable data 12. A Spanish observational study of 755 adults with type 1 diabetes using the FreeStyle Libre 2 found that 248 of them (32.8%) had at least one night in two weeks when a low below 3.9 mmol/L (70 mg/dL) was followed by a high above 10.0 mmol/L (180 mg/dL) before 6 a.m. 13. The authors could not tell whether those highs came from carbohydrate eaten to treat the low, from hormones, or from both 13.
Taken together, a night-time low is a less common explanation for a high waking number, though it can happen, particularly after an over-treated low 14. Sensors can also mislead at night. In healthy volunteers filmed while asleep, sudden drops in sensor readings lined up with times they lay on the sensor, probably because pressure reduced local blood flow 15. A short, sharp overnight dip deserves a second look before it is counted as a real low.
The same 7 a.m. number can come from very different nights. These are the main alternatives to rule in or out with your care team.
Which option fits depends on the cause, the type of diabetes and your current treatment. These approaches come from guidelines and studies, for discussion with your team. None is a reason to change a dose on your own.
Because evening NPH peaks early and then wanes 7, the type and timing of background insulin both matter. For type 2 diabetes, the review reports that basal insulin given in the evening abolishes the dawn rise by restraining the liver’s glucose output 7. Any change has to balance the early-morning rise against the risk of lows earlier in the night, when insulin can run high relative to need 3, 8.
Diabetes Canada’s 2025 type 1 guideline notes that insulin pumps allow finer adjustment than injections for people whose basal needs differ across the day, and it names the dawn phenomenon as an example 17. The same guideline recommends automated insulin delivery (AID) systems as the preferred treatment for people with type 1 diabetes who are willing and able to use them. These systems link a pump to a CGM and use an algorithm to automate parts of insulin delivery 17.
Oral medications of the time did not eliminate the dawn rise in the CGM study 2, and the 2013 review reported that they do not adequately control it, even in combination 7. Medication choices depend on your whole treatment plan, so this is a question for your prescriber.
In a small randomized trial, 12 adults with type 2 diabetes each completed two of three seven-hour sitting sessions, broken up by three-minute light walks every 60, 30 or 15 minutes 18. With walks every 15 minutes, fasting glucose the next morning fell by about 1.0 mmol/L (18 mg/dL), compared with almost no change with hourly breaks, and the dawn rise was shorter and smaller than with less frequent breaks 18. The study was small, short and tested daytime breaks, so it shows what is possible more than what to expect. Dinner is another lever to review with your team, given how long a high-fat meal can keep glucose up in type 1 diabetes 16. That study tested fat content only, so the effect of dinner timing on its own is less clear.
Because the dawn rise changes from night to night 1, one night tells you little. Over 7 to 14 nights, note the following.
Then look at the shape of each night.
| Overnight shape | What it can suggest |
|---|---|
| Steady until the early hours, then a climb to waking | Dawn phenomenon 1, 8 |
| High at bedtime and stays high | Effects of dinner, a snack or evening insulin 16 |
| Slow, steady rise from bedtime on | Background insulin fading 7, 9 |
| Low, then high before morning | Over-treated low or a rebound 11, 13, 14 |
| Brief, sharp dip while asleep | Possible sensor pressure artifact 15 |
These are starting points for a conversation with your care team, who will read them alongside your treatment and history.
It may be the dawn phenomenon, an early-morning rise in glucose without a low beforehand 1. It can also come from dinner, a bedtime snack, background insulin that fades before morning, or a night-time low treated with extra carbohydrate. Tracking 7 to 14 nights helps your care team tell these apart.
Toward dawn, the liver releases more glucose while tissues take up less, for the same amount of insulin 6. Night-time growth hormone surges are the best-supported trigger. In six people with type 1 diabetes studied in 1985, suppressing them stopped the rise, which returned when growth hormone was added back 6.
Researchers defined it as an abrupt increase in fasting glucose, insulin needs or both between about 5 and 9 a.m., with no low beforehand 1. In a study of 114 people with type 1 diabetes, insulin needs fell to a low point around 2:40 a.m. and peaked around 6:40 a.m 8.
Yes. In a 1984 study of eight volunteers without diabetes, glucose and insulin rose after about 5:30 a.m. as the pancreas secreted more insulin to match 5. In type 1 diabetes, and in type 2 when the body's own insulin response is reduced, that automatic answer is weak or missing.
The dawn phenomenon is an early-morning rise in glucose without a low beforehand 1. The Somogyi effect is the idea that a night-time low triggers insulin-opposing hormones that drive glucose high by morning 9. A night-time low is a less common explanation, though it can happen, particularly after an over-treated low 14.
Look at 7 to 14 mornings in a row. In Glyve’s day view, the glucose line shows your own target range shaded, with logged meals, insulin, exercise, sleep and stress placed on the line, so you can see whether a climb starts after midnight or soon after dinner. Logging dinner by voice or photo makes the evening easier to review, and each estimate is a draft you check before saving. The week view shows whether the pattern repeats. Glyve reads glucose from Apple Health, and Dexcom and FreeStyle Libre readings through those companies’ sharing services, which can arrive later than in your CGM app, so check that the night has filled in before you judge it. Glyve shows what happened together; it doesn’t tell you what caused it. A PDF report or CSV export lets you share the two weeks with your care team.
How we write about diabetes. Spotted a mistake? Email info@glyve.ca and we’ll check it against the source.
Glyve keeps glucose, food, insulin, exercise, sleep and stress on one timeline.
For iPhone, iPad and Apple Watch.