Exercise and Type 1 Diabetes

Start here if exercise feels daunting

If formal exercise management feels like a lot, the most accessible entry point is activity snacking: ten to twenty minutes of movement after the meal that produces your biggest spike. It is lower risk than structured exercise and it has its own dedicated guide.

Read the Activity Snacking guide

Major in the majors

Three variables decide almost everything

If exercise has felt chaotic, start here. Across almost all forms of exercise, glucose behaviour is dominated by three variables, in order of importance.

Insulin on board is also the variable you can most directly change. When these three are prioritised first, most exercise-related hypoglycaemia and volatility becomes predictable from the starting conditions.

The core problem exercise creates

Exercise is hard to manage in type 1 diabetes not because it is dangerous, but because it amplifies insulin action.

Muscle contraction increases glucose uptake. Blood flow accelerates insulin delivery. Counter-regulatory hormones may push glucose out of the liver. These processes act quickly. Injected or pumped insulin does not.

Why exercise hypos are usually insulin problems, not exercise problems

Large glucose drops during exercise are rarely caused by exercise itself. They are almost always caused by supercharged insulin action: exercise increasing blood flow and glucose uptake, amplifying the effect of insulin that is already present. This is physiology doing exactly what it should.

The implication is practical: reducing insulin on board before exercise, through timing, bolus reduction, or basal adjustment on a pump, addresses the actual cause rather than the symptom.

Counter-regulation is real, but blunted in T1D

In people without diabetes, falling glucose during exercise triggers a prompt counter-regulatory response (glucagon and adrenaline release glucose from the liver). In long-standing type 1 diabetes this response is often blunted, which is part of why the same session that a friend without diabetes shrugs off can drop you hard. The muscle also acts as an endocrine organ: contracting muscle releases myokines and increases insulin-independent glucose uptake, which continues for hours after the session ends.

Risk factor 1, the one that matters most

How much bolus insulin is on board

Because exercise amplifies insulin action, the single most important thing to know before you start is how much bolus insulin is still working. This is the main reason to try to eat and give insulin about three hours before a planned session, or to exercise first thing in the morning before any meal bolus: both leave you with low insulin on board and a far more predictable session.

A simple, robust way to think about it:

• Bolus within the last 3 hours: insulin on board is meaningful. Hypo risk during exercise is high. Plan a reduction or carbohydrate.
• Bolus more than 3 hours ago (or fasted / basal only): insulin on board has fallen. Hypo risk is reduced, and the session becomes far more predictable.

The 90-minute fuel window

If you have to eat close to exercise, aim within 90 minutes

When exercise falls within roughly 90 minutes of a meal bolus, insulin action is often near its peak, and a bolus reduction is commonly needed to avoid a predictable hypo. Exercise and peak bolus action overlap.

The practical consequence: if you are going to eat within about two hours of exercise, try to make it within 90 minutes of starting, so you can begin the session having already reduced the bolus, rather than trying to catch a fall with carbohydrate mid-session. Smaller, earlier interventions consistently outperform larger, later corrections. Even an AID algorithm cannot override the physiology of this window; it can only react to the glucose that results.

Setting the starting insulin conditions

The reduction protocols

Two population-average starting frameworks: 50/50/20 for multiple daily injections, and T25/T25 for an automated insulin delivery system (hybrid closed loop). Both are teaching simplifications, not fixed doses; any insulin change is a care-team decision.

50/50/20: for MDI (multiple daily injections)

For more specific guidance than the general 50/50/20 starting point, GNL’s Exercise Planner uses a graded reduction table by exercise type and duration.

GNL’s operational reduction table for those wanting more specifics than the 50/50/20 starting point, informed by Rabasa-Lhoret (2001) and the EASD/ISPAD 2025 consensus (Moser, Zaharieva, Pemberton); not a fixed prescription, agree changes with your care team. AID users have a simpler in-built gate and do not need this table.

T25/T25: for AID (automated insulin delivery / hybrid closed loop)

With the pre-exercise insulin conditions set, the starting point is done. The next thing to read, before and throughout the session, is glucose and its trend.

Risk factor 2, read it before and every 30 minutes

Starting glucose and the trend arrow

Once insulin on board is accounted for, the next question is where your glucose is and which way it is moving. This is not a one-off check at the start; it is a running assessment before you begin and roughly every 30 minutes during the session. A number without a direction is incomplete: 7 mmol/L falling fast and 7 mmol/L rising slowly are two different situations.

A weight-and-glucose starting guide for carbohydrate

Body weight and starting glucose set the starting carbohydrate for a 30-minute session, with a meaningful recent bolus on board. The rates below are the clinical per-kilogram figures used on this guide: 0.5 g/kg at 4 to 7 mmol/L, 0.3 g/kg at 7 to 10 mmol/L, and 0 g/kg at 10 to 15 mmol/L and above (the higher glucose already provides the headroom). Weight is capped at 60 kg, the population-average ceiling used across every GNL carb calculator. It is an education tool, not a personal dose.

Below 4 mmol/L: treat the low first and do not start, using the Hypo Treatment approach. At 15 mmol/L or above: hold, and check for ketones before any activity. No single 30-minute window needs more than 30 g, because the gut absorbs glucose at roughly 1 gram per minute.

Adjusting for the trend arrow

Take the figure from the table, then adjust it for the direction of travel (cap the result at 30 g). Worked example, a 40 kg person in the 4 to 7 mmol/L band, starting figure 20 g:

Clinical per-kilogram rates (0.5 / 0.3 / 0 g/kg), aerobic exercise, with the 60 kg weight cap and 30 g absorption ceiling per GNL calculator canon. See the note above on how these relate to the live explorer.

Different kinds of exercise behave differently

Endurance, mixed, anaerobic and competition

The type of exercise shapes the direction glucose tends to move. The live explorers use three categories; competition sits on top as a framing overlay rather than a fourth setting.

• Endurance / aerobic (steady running, cycling, swimming): tends to lower glucose, sometimes sharply. The classic hypo-risk profile.
• Mixed (team sports, circuits, interval work): a tug-of-war. Aerobic segments pull glucose down; short intense bursts and adrenaline can push it up. The net effect is less predictable.
• Anaerobic (sprints, heavy resistance work, HIIT): adrenaline and hepatic glucose output can drive glucose up during and just after the session, with a delayed fall later.
• Competition: not a separate physiology, but a framing overlay. Adrenaline runs higher, the day is less controllable, and the stakes of a hypo are greater. This is exactly where leaning on the Three-Hour Rule (starting with low insulin on board) earns its keep.

How much the type matters depends heavily on insulin on board. If insulin on board is high, glucose will tend to fall whatever the activity type; the circulating insulin dominates. If you are in basal-only conditions (three or more hours after a meal bolus), the type of exercise then has much more influence, and it is what should shape the plan. An AID algorithm sees only the resulting glucose, not whether the effort was aerobic or anaerobic, so anticipating the type is still your job.

New to this? Do not try to model all of it at once. Start aerobic, start with the Three-Hour Rule where you can, and use 50/50/20 (or T25/T25 on a loop) as your default starting point. Refine from there with your CGM data over weeks, not within a single session.

After the event

The post-exercise window is the harder problem

The session is not where planning ends. Insulin sensitivity stays elevated for hours afterwards, and the one-to-two hour window after exercise, plus the following evening and overnight, is where a lot of exercise hypos actually happen, even when the session itself went smoothly.

This is the second half of both frameworks:

• 50/50/20 (injections or pump): cut the next meal bolus after exercise by about 50 percent, because sensitivity is still high. For exercise in the evening (after about 4 pm), use any one of a 20 percent basal reduction overnight, 20 g of carbohydrate without insulin, or 20 g of protein, to protect against a later or overnight low.
• T25/T25 (closed loop): return the target to normal once the session is done, then cut the next meal bolus by about 25 to 33 percent. The loop keeps helping in the background, which is why the manual cut is smaller.

Two patterns worth knowing: late-afternoon vigorous exercise tends to shift hypo risk into the evening rather than overnight, and long sessions (over about an hour) can leave elevated hypo risk into the following day. Smaller, earlier interventions again beat larger, later corrections.

Ten tactics that hold up in real life

Key tips for exercise

1. Treat CGM accuracy during exercise as conditional. CGM is your best learning tool, but it is less accurate during exercise. Read the number alongside the trend arrow and the context, and be slower to react to a single mid-session reading. Iterate over weeks, not within a session.
2. Run the three-majors checklist before every session: insulin on board (how much, and when was the last bolus), starting glucose, and trend. Get those right and the rest is refinement, not rescue.
3. Sex differences are real and underappreciated. The same session on different days of the menstrual cycle can behave quite differently. That is signal, not noise; tracking it over a few cycles often reveals the pattern.
4. Fear of hypoglycaemia is a planning variable. The fear is reasonable; the cost of avoiding exercise is usually larger than the hypo it prevents. A rehearsed hypo response, fast carbs within reach, a plan made with your team, and sensible CGM alarms all lower the cost of starting.
5. Barriers and facilitators matter as much as physiology. The right plan is one that fits the rest of your life. The most evidence-based session you cannot actually do is worse than a less optimal one you can repeat for years.
6. Altitude changes the rules. The first few days at altitude often need different settings from sea level. Plan altitude trips with your diabetes team rather than improvising on arrival.
7. Muscle damage matters for days, not hours. Heavy eccentric or novel work (long downhill runs, new resistance training, certain team sports) can alter insulin sensitivity for two or three days afterwards, not just in the immediate post-exercise window.
8. Calibrate to real-world data, not the trial average. Large datasets (T1DEXI and its follow-ons) show hypo risk on exercise days is meaningfully higher than on non-exercise days, and that real-world response is more variable than pivotal trials suggest. The trial average is a starting point, not a forecast for you.
9. What the largest GNL-analysed datasets show. Exercise improves time-in-range (modestly, a little more for AID users); the biggest gain from daily steps is between zero and about five thousand steps; strength training appears the safest type for hypo risk; long sessions leave elevated risk into the next day; and on AID, more frequent correction boluses did not raise time-below-range, the automated suspension does its job.
10. Know when to ask your care team. This guide is education, not prescription. Bring your team in for recurrent unexplained exercise hypos, post-exercise highs that do not resolve, new exercise types or intensities that shift the pattern, pregnancy, altitude trips, and any change to your insulin regimen or AID system. The point of this guide is to make that conversation a structured one.

Before you start, a short safety checklist

Set up for a safe session

Set your CGM low alert at 5.0 mmol/L

CGM lag lengthens during exercise: the sensor can read meaningfully higher than your true glucose when glucose is falling, and the delay stretches from the usual few minutes to around 12 minutes during prolonged aerobic work. A low alert set at 5.0 mmol/L gives you the headroom to act before you are actually low. If a hypo is suspected mid-session, confirm with a fingerstick rather than trusting the sensor alone.

On an AID system, two ways to set stable conditions

The mainstream default (stay in closed loop): switch on your system’s exercise feature 60 to 90 minutes before you start and leave it running through the session. On Tandem Control-IQ this is Exercise Activity mode (raises the target to 8.9 mmol/L); on MiniMed 780G it is Temp Target (raises the target to 8.3 mmol/L). Both give the algorithm time to reduce insulin on board before you begin, while keeping the automated safety net on.

Have plenty of hypo treatment to hand

Fast-acting carbohydrate within reach is non-negotiable for any structured session. Rehearse the response so it is automatic, not a scramble.

Review and adapt

Treat every session as data. Look back at the CGM trace, note what your starting conditions were, and adjust the next attempt. Exercise with type 1 diabetes is learned over weeks and months of iteration, refined with your diabetes team, not solved in one session.