Read this before you use any formula on this page
The formulae below produce starting estimates — the same ones clinicians use to open a conversation, not to close one. Your real ratios come from your own data and your doctor’s judgement, and they will almost certainly differ from the calculation. Never change an insulin dose on the strength of an article. Use this to understand what your team is doing and to ask better questions at your next review.
Everything begins with your total daily dose
Your TDD is every unit of insulin you take in an average 24 hours — basal plus all your boluses, added together. It is the single input both of the classic ratio formulae depend on, which is why getting it right matters more than the arithmetic that follows.
Take a genuinely representative week rather than yesterday. Add up each day’s total and average the seven. Skip days that were obviously atypical — a stomach bug, a wedding, a day you were unwell or ran a half marathon. If you are on a pump, the device reports this for you; on MDI, your pen’s memory or a fortnight of logbook entries will do it.
A high TDD makes the formulae less trustworthy
These rules were derived in adults with reasonably typical insulin sensitivity. If you are heavily insulin resistant, in puberty, pregnant, or on steroids, your TDD is inflated for reasons the formula knows nothing about — and it will hand you ratios that are far too aggressive. The rules are a first guess for a stable adult, not a universal law.
Work out your starting numbers
Enter your TDD and a hypothetical meal. Everything updates live. Nothing is stored or sent anywhere — the maths runs entirely in your browser.
Estimated total for this meal: 5.2 units
That is the meal dose plus the correction dose, before accounting for any insulin still active from an earlier bolus. See the section on stacking below.
ICR — how much carb one unit covers
Your insulin-to-carb ratio answers one question: how many grams of carbohydrate does a single unit of rapid insulin cover? The conventional starting estimate is the 500 rule.
ICR = 500 ÷ TDD. A TDD of 20 units gives an ICR of 25 — one unit for every 25 g of carbohydrate. A TDD of 50 gives an ICR of 10, because a more insulin-resistant body needs more insulin per gram.
Written as a ratio you will often see it as 1:25 or 1:10. To dose a meal you divide: 60 g of carbohydrate at an ICR of 1:25 is 60 ÷ 25 = 2.4 units.
Two things trip people up here. First, most people do not have one ICR — insulin resistance is highest in the morning, so a common pattern is a tighter ratio at breakfast (say 1:8) and a looser one at dinner (1:12). Second, the ratio is only as good as your carb counting. If you are consistently 20 g out on your estimate of a plate of rice, no ratio in the world will save the reading, and chasing the ratio will only make things worse.
Fix the carb count before you touch the ratio
Weigh your five or six most repeated meals once, properly, with a kitchen scale. A standard katori of cooked rice, two chapatis, your usual bowl of poha. You do not need to weigh forever — you need an accurate anchor for the foods that make up most of your week, so that a surprising reading tells you something about your insulin rather than about your guesswork.
ISF — how far one unit drops you
Your insulin sensitivity factor, also called the correction factor, is the number of mg/dL that one unit of rapid insulin is expected to lower your glucose.
ISF = 1700 ÷ TDD for rapid-acting analogues. A TDD of 20 gives an ISF of 85 — one unit is expected to bring you down about 85 mg/dL.
| Your insulin | Constant to use | Why |
|---|---|---|
| Lispro, Aspart, Glulisine | 1700 (some clinicians use 1800) | Rapid analogues act faster and clear sooner |
| Regular / Actrapid / Human R | 1500 | Slower onset, longer tail — a unit does less per hour but works longer |
| Working in mmol/L | 100 ÷ TDD | Same rule, converted — the 1700 figure is mg/dL only |
You will see 1800 quoted as often as 1700, and the honest answer is that the choice of constant matters far less than what your own readings tell you. Start with whichever your team uses, then adjust from evidence.
The correction dose formula
Once you have an ISF, correcting a high reading is a single line of arithmetic: how far above target you are, divided by how far one unit takes you.
(Current glucose − Target glucose) ÷ ISF. At 400 mg/dL with a target of 100 and an ISF of 85: (400 − 100) ÷ 85 = 3.5 units.
Your target is a number your doctor sets, not one you pick from an article. It is usually higher than you would expect — commonly around 100 to 120 mg/dL, and deliberately higher again if you live alone, have hypo unawareness, are elderly, or are about to drive or exercise. A target that is too tight is the most common cause of self-inflicted lows.
Where corrections fit against the usual adult ranges
A worked example, start to finish
Meet a hypothetical adult with a TDD of 34 units, on Aspart, sitting down to dinner.
Establish the ratios
ICR = 500 ÷ 34 ≈ 1:15. ISF = 1700 ÷ 34 = 50. So one unit covers roughly 15 g of carbohydrate and is expected to drop glucose by about 50 mg/dL.
Count the meal
Two chapatis (about 30 g), a katori of dal (about 15 g), sabzi and curd (about 5 g) comes to roughly 50 g. Meal dose = 50 ÷ 15 = 3.3 units.
Add the correction
Pre-meal reading is 210 mg/dL against a target of 110. Correction = (210 − 110) ÷ 50 = 2 units. Running total: 3.3 + 2 = 5.3 units.
Subtract insulin still working
A 2-unit correction taken 90 minutes ago is likely only half spent, so roughly 1 unit is still active. Deduct it: about 4.3 units. This step is the one people skip, and it is the one that causes lows two hours later.
Adjust for context, then dose
A long walk after dinner, alcohol with the meal, or a heavily fatty dish all change the answer. Context is judgement, not arithmetic — and it is where an experienced person with diabetes beats any formula.
Insulin on board, and why stacking hurts
Rapid insulin does not finish when the spike does. It keeps working for roughly four to five hours. Insulin on board — IOB — is the portion of your earlier doses still active.
Stacking is what happens when you correct a high, feel impatient after an hour, see it is still high, and correct again. The first dose had not finished. Both doses land together, and the high becomes a low — usually a bad one, usually at an inconvenient time. The impatience is completely understandable and it is still the single most common self-inflicted hypo in insulin users.
Wait the full interval before correcting again
As a rule of thumb, give a correction three to four hours before deciding it did not work. If you use a pump or a smart pen, the device tracks IOB and subtracts it for you. On pens and syringes you have to do it yourself: note the time and the size of every correction, and assume roughly half of it is still active at the two-hour mark.
How to tell which number is wrong
When readings go wrong, people tend to blame whichever ratio they last heard about. The pattern of the error tells you which one is actually at fault.
| What you see | Likely culprit | What to look at |
|---|---|---|
| Drifting up or down overnight, with no food involved | Basal dose | Basal first |
| Flat before the meal, high 3–4 hours after | ICR too loose | Meal ratio |
| Low 3–4 hours after eating, from a normal start | ICR too tight | Meal ratio |
| Corrections consistently undershoot target | ISF number too large | Correction factor |
| Corrections consistently overshoot into a low | ISF number too small | Correction factor |
| Big spike at 1 hour that settles by 3 | Bolus timing, not the ratio | Pre-bolus earlier |
Basal comes first, always
If your basal is wrong, every meal ratio you calculate on top of it is being fitted to a moving floor, and you will chase your own tail for months. Sort the basal with your doctor first — that is what basal testing is for — and only then tune ICR and ISF. Change one variable at a time and give each change three to four days before judging it.
Your ratios are not permanent
A ratio that was right in March can be wrong in July. This is not failure — it is how the condition behaves, and knowing the usual suspects saves you from concluding you are doing something wrong.
Push you towards needing more insulin
Ratios tighten, ISF number falls
- Illness and infection — often dramatically, for a few days
- Steroids — including a short course for a chest infection
- Puberty and pregnancy — sustained, and substantial
- The week before a period — for many, predictably
- Stress and poor sleep — cortisol raises resistance
- Weight gain and inactivity — gradual but real
Push you towards needing less
Ratios loosen, ISF number rises
- Regular exercise — with an effect lasting well past the session
- Alcohol — the liver stops releasing glucose; lows come later
- Weight loss — sensitivity improves as it comes off
- Hot weather — absorption speeds up in the heat
- The honeymoon phase — early type 1, residual insulin production
- Kidney impairment — insulin clears more slowly; needs review
The Indian summer effect is real
Insulin absorbs faster from warm skin, and the difference between a February and a May afternoon is enough to change how a dose behaves. It is also worth checking your storage: insulin left in a car or near a window in a 45°C summer degrades, and unexplained highs across a whole vial are more often a storage problem than a ratio problem.
What to bring to your endocrinologist
A ratio conversation goes far better with evidence than with a description. Fourteen days of honest data will get you further than an hour of recollection.
Your CGM export or two weeks of meter readings, your current basal and bolus doses, the ratios you are using now, and a note of any hypos with the time and likely cause.
Any pattern you have noticed by time of day — the 4 a.m. rise, the post-breakfast spike, the Tuesday-after-football low. Patterns are what get ratios changed.
Whether you should have different ratios by meal, what your correction target should be, and how long to wait between corrections given your insulin.
One change, not five, and a date to review it. Multiple simultaneous changes make the result uninterpretable.
Better data makes better ratios
Ratio tuning is guesswork without continuous readings behind it. Browse CGM sensors, meters and strips with free pan-India shipping — and see our CGM comparison if you are deciding between systems.
The bottom line
500 over TDD, 1700 over TDD, and glucose minus target over ISF. Three lines of arithmetic that give you a defensible starting point and a shared language with your doctor. What makes them work is not the formula but the follow-up: change one thing, wait a few days, read the pattern, and let the evidence — not the calculator — settle where your numbers land.
Keep reading
See all guides →New
Treating a hypo — what actually works, and what wastes time
Test, treat, retest. What 15 g of fast carbs looks like in India, why chocolate is the wrong tool, glucagon and severe hypos, night lows and hypo unawareness.
New
Sick day rules — managing diabetes when you're ill
Why basal insulin never stops, how often to test, ketone thresholds and what each level means, correcting with bolus not basal, and the signs that mean hospital.
Guide
The complete guide to travelling with diabetes
Packing checklists, glucose targets, airport security tips and country-by-country emergency phrases for diabetics travelling from India.