- The Child Maintenance Service (CMS) counts overnights — the nights your child sleeps in your care — and sorts them into bands: 52, 104, 156 and 175+ nights a year. Each band reduces the maintenance calculation by a set fraction. (How the underlying amount is worked out — gross income, the rates, what pension contributions do to it — is its own subject and deserves its own guide; this one stays on the nights.)
- If care is exactly equal, no maintenance is payable for that child. That's the territory of Regulation 50.
- If you each tell the CMS a different number of nights and neither of you can prove yours, the CMS falls back on low defaults: it can assume shared care of just one night a week — or, where you both agree it's "at least 52 nights" but not on the exact figure, it uses 52. In other words: no evidence, bottom of the range.
- Which means the practical answer to "how do I prove my nights?" is: a truthful record, kept nightly, as it happens. Not photos of children in beds. Not memory. A boring, dated, unedited log — corroborated by court orders, independent records like school and GP correspondence, and the evidence you already have without realising: the shared family calendar, the messages between you agreeing each week's plan. (The kitchen blackboard wipes clean, sadly — if that's how your family plans, photograph it now and then; the photo carries the date.)
- You can start that record tonight. It gets stronger every day you keep it.
If you've landed here mid-dispute — because the CMS has been given a picture of your week that you don't recognise, and you know the children's shoes by the door tell a different story — hold on. The mechanics are on your side more than the letter suggests. The rules are more mechanical than they feel right now, and mechanical is good news: mechanical things can be evidenced.
How does the CMS count shared care?
In CMS language, shared care means your child stays overnight with the paying parent — the parent who doesn't have the main day-to-day care (the law calls them the "non-resident parent", even when that label feels absurd). The other parent is the receiving parent.
The CMS works out a weekly maintenance figure from the paying parent's gross income, then reduces it according to how many nights a year the child is expected to stay with the paying parent. The bands, as published on GOV.UK, are:
| Nights of shared care per year | Reduction for that child |
|---|---|
| 52 to 103 | 1/7th (14.29%) |
| 104 to 155 | 2/7ths (28.57%) |
| 156 to 174 | 3/7ths (42.86%) |
| 175 or more | Half (50%), plus an extra £7 a week |
Two details worth knowing:
- The reductions have a floor. GOV.UK puts it bluntly: for parents on the reduced or basic rate, "the child maintenance payments cannot go below £7 a week". So shared care can shrink the bill a long way — even a parent in the top band, with the children more nights than not, still pays at least that token amount. The only ways to zero are establishing equal care (the Regulation 50 territory) or one narrow exception: a paying parent on benefits, paying the £7 flat rate, whose child stays 52 or more nights a year, pays nothing for that child.
- The bands only matter at their edges. Moving from 140 to 150 nights changes nothing; moving from 140 to 156 changes the band, and either parent can report it.
How does the CMS decide which number to use? Regulation 46 of the Child Support Maintenance Calculation Regulations 2012 sets it out: the CMS looks at the number of nights the paying parent is expected to have the child overnight during the next 12 months. To decide that, it must consider any court order or agreement between you — and if there's neither, whether a pattern of shared care has already been established over the past 12 months. Hold onto that phrase. Your record is how a pattern gets established.
"an established pattern" — that's your record's job.
A night counts when you have care of the child overnight and the child stays at the same address as you. (There's even a rule for boarding school and hospital stays: the parent who would otherwise have had the child that night is treated as having care.)
And at exactly equal? The GOV.UK factsheet is plain: where day-to-day care of a child is equal between the two parents, the paying parent doesn't have to pay any child maintenance for that child. Proving equal care — rather than 175+ nights — is its own subject, and it's covered in our Regulation 50 guide.
What happens when the CMS is told two different numbers of nights?
This is the part nobody warns you about, and it's the reason this guide exists.
When parents disagree, the CMS asks both of you for evidence — it will take "formal, informal or spoken evidence" of any agreement about shared care, and it will accept a court order setting out the arrangements. But here's what the published rules say happens when the disagreement can't be resolved:
- If you both agree care is shared but can't agree the number of nights, the CMS may apply "assumed shared care" — an assumption of just one night a week, giving the smallest reduction, 1/7th. That decision then stands until you reach an agreement or one of you provides proof of a different number.
- If you both agree it's at least 52 nights but can't agree how many more, GOV.UK states the CMS "will assume that shared care is only happening for 52 nights a year" — again, the bottom band.
Read those defaults again slowly, because they decide disputes. When neither side can prove their number, the calculation doesn't settle in the middle. It settles low. A parent doing three nights a week who can't evidence it can end up assessed as if they do one. This is what parents on every forum discover the hard way, and it's not a conspiracy — it's just what the published fallback rules do when no evidence arrives.
Parliament has noticed. In a Westminster Hall debate on the Child Maintenance Service in March 2026, the system was described as one that:
Hansard · Westminster Hall · 17 Mar 2026"…accepts unverified claims, but refuses genuine evidence."
I learned this the expensive way: doing half the care and still being assessed at over £700 a month, because at the moment it mattered, the account of my nights the CMS worked from wasn't mine. The rules don't reward the parent who does the most care. They reward the parent who can show it. Which brings us to evidence.
What evidence does the CMS accept for shared care?
Nothing in the published rules gives you a tick-box list, but the regulation tells you the order things are considered in, and experience fills in the rest. In rough order of weight:
1. Court orders and formal agreements. Regulation 46 requires the CMS to consider "the terms of any agreement made between the parties or of any court order providing for contact" first. If a court order sets out your nights, that's the strongest single document you can hold. A written parenting agreement — even an informal one, though one drawn up with a solicitor carries more — comes next. If you have either, the number in it is your starting point.
2. Independent third-party records. Where there's no order or agreement, the question becomes the established pattern over the past 12 months — and the most persuasive picture of a pattern is one drawn by people with no stake in your dispute. School correspondence addressed to you at your address. GP registration and appointment letters. Nursery and club registers showing who drops off and collects. None of these count nights directly, but together they corroborate that your home is one of the child's homes, week in, week out. A point that trips people up: "independent" is about who created the document, not who keeps it. The school letter sitting in your kitchen drawer is still the school's word — scan it, file it, it counts. But material you create yourself — photos of yourself at drop-off, at sports day, at the school play — isn't independent, however real the moment was. It has a place, and that place is next.
3. Everything you create yourself — the diary, and the material behind it. Your own record lives or dies by two principles. First: contemporaneous beats reconstructed. "Contemporaneous" just means written down at the time — one dull dated line a night — and it's the whole difference between a record and a recollection; an account assembled months later, in one sitting, in one pen, persuades nobody, and shouldn't. Second: supporting material corroborates — it never carries. Receipts, handover messages and photos sit behind a proper nightly record and agree with it; on their own, parents report they get short shrift ("CMS will not accept photos as evidence", as one Money Saving Expert thread puts it flatly). How each evidence type ranks in detail, category by category — with templates for capturing each one — is exactly the job of the Reg 50 Playbook.
One thing said plainly, because our whole approach depends on it: the only record worth keeping is a true one. Don't backdate, don't pad, don't tidy history into what it should have been. A truthful record survives cross-examination; an embellished one collapses and takes your credibility down with it. If your real number is 46%, a good record will say 46% — and you'd still rather know that from your own diary than hear it from someone else's.
How do you keep a record that stands up?
Here's the whole method. It isn't clever, which is exactly why it works.
Log every night, as it happens. One line a day: the date, where your child slept, and one ordinary detail. Thirty seconds before you go to bed. The entry made tonight about tonight is worth ten entries made in December about June.
Count the way the CMS counts. The CMS thinks in annualised overnights — nights per year, sorted into bands. So keep a running count: nights in your care over the last 12 months, divided by the total. Watch which band you're actually in, not the band you feel you're in. A "roughly every other week plus Wednesdays" arrangement can sit anywhere from 130 to 180 nights depending on holidays — you want to know, precisely, before anyone else claims to.
Record the deviations honestly. The missed night when your child was ill at the other house. The holiday fortnight. The swapped weekend. These go in the log the same as everything else, because a record with honest gaps is believable and a record without any isn't. Term-time and school-holiday patterns usually differ; a full year of logging averages them for you, the same way the 12-month rule does.
I did this for 447 consecutive days. The result was a split my record could state precisely — around 54% of overnights, and just under 52% of care hours over the first full year — backed by an evidence bundle that grew past 200 files. But the number was never really the point. The point was that nobody could argue with how it was kept: nightly, dated, unedited, dull as ditchwater. When you can put that on the table, the conversation stops being your word against theirs. There's only one record in the room.
Which mistakes make a care record easy to dismiss?
Five, mostly — and I understand the pull of every one of them:
- Reconstructing from memory. A calendar filled in retrospectively looks like what it is. Start from today; don't fake yesterday.
- Suspiciously round numbers. "It's exactly 50/50" and "always 3 nights a week" read as estimates. Real care is lumpy — 51.6% is more believable than 50% precisely because it's less tidy.
- Logging only your good weeks. A record that skips the fortnight your child stayed away isn't a record, it's an argument. Log the weeks that don't help you; they're what make the rest count.
- Turning the diary into a grievance file. The moment your log starts scoring points against the other parent, it stops being evidence of care and starts being evidence of conflict. Keep it about your child and your nights. Nothing else belongs in it.
- Arguing money instead of nights. Bank statements showing you bought the school shoes feel like proof of care — but the question the CMS is asking is where the children slept, not who spent more. "I pay for everything" doesn't move the count by a single night, and it slides the record towards the grievance file above. Spending has a quieter, better use: a supermarket receipt near your home on your night corroborates presence. As backing for the nightly record, useful; as a rival argument to it, worthless.
FAQ
How do I prove my children stay at home with me — do I send pictures of them in their beds?
No — and parents who've tried report that photos get rejected. What the rules actually respond to is an order or agreement stating nights, or an established 12-month pattern. A nightly dated log plus independent records (school, GP, clubs) is how a pattern gets shown. Photos can sit behind that record; they can't replace it.
The other parent has told the CMS the children only stay a handful of nights a year. What can I do?
The CMS should ask both parents for evidence of their number. If neither claim can be evidenced, the published defaults are low — one night a week, or 52 nights. So the practical response is to put your evidence in: any order or written agreement, corroborating records, and your log of actual nights. And if you haven't been logging: start tonight, because the pattern over the coming months can still be established and shown.
Will the CMS accept location screenshots, receipts or statements from neighbours?
Nothing published rules them out, but forum experience is that loose material on its own carries little weight. The rules are built around agreements, orders and established patterns — so treat this kind of material as corroboration for a proper nightly record, not a substitute for one.
We genuinely share care equally — why am I assessed to pay anything at all?
Because equal care has to be established, not assumed — and until it is, one of you is treated as the paying parent (usually the one not receiving Child Benefit). Where equal day-to-day care is established, no maintenance is payable for that child. How that works, and how to rebut the assumption, is the subject of our Regulation 50 guide.
Do school holidays change my count of nights?
They're part of it. The CMS looks at expected nights across a full 12 months, so term-time weeks and holiday weeks average together into one annual figure. That's another reason a full year of honest logging beats any estimate: it does the averaging with real dates, including the holidays that didn't go to plan.
Sources
- How the Child Maintenance Service works out child maintenance — GOV.UK (Step 6: shared care; night bands and reductions; the 52-night assumption; the £7 floor)
- How your child's living arrangements affect child maintenance — DWP factsheet, GOV.UK (updated 27 October 2023: band fractions, assumed shared care, evidence, equal-care rule)
- The Child Support Maintenance Calculation Regulations 2012, regulation 46 — legislation.gov.uk (how the number of nights is determined; the 12-month pattern; what counts as a night)
- Child Maintenance Service — Westminster Hall debate, 17 March 2026 — UK Parliament, Hansard
Figures and rules above reflect the GOV.UK guidance and legislation as checked in August 2026. The rules can change — rely on the "last updated" date at the top of this page.
You've just read the rules. Keeping the record is the other half — one honest line a night, filed and counted the way the CMS counts.