One in ten children looked after in England on 31 March 2025 had three or more placements during that year, and the rate is highest among those looked after continuously for less than twelve months

In these statistics a placement is a recorded care arrangement. It usually corresponds to where a child lives, but a new placement can also be recorded when the child stays with the same carer and the status of the arrangement changes. When the Department for Education reports high placement instability, it means a child who had three or more placements during a single year to 31 March. That threshold is an administrative line, not a description of a childhood.

The difference between a placement and a home is not pedantic. The return local authorities make starts a new episode whenever a child's placement changes, and its code set separates a change of placement and carer from a change of placement where the carer stays the same. "Change in the status of a placement only" is one of the reasons recorded for the second kind: the child stays where they are while something about the arrangement changes, such as a foster carer moving from a council to an agency. In the year to March 2025 that reason was recorded against 10,500 placement changes, 17.8% of the total.

So a child counted as having three placements has not necessarily moved home three times. This report says placements throughout.

A move is not automatically a harm either. Children move to go home to their families, to move in with adopters, to join a brother or sister, or to step towards living independently. The largest recorded reason category for a placement change in England is "change to/implementation of the care plan". Equally, a child who has not moved has not necessarily been well looked after: staying put in a placement that is wrong for you is not a good outcome, and children in the published research describe asking to move as well as dreading it.

What the count captures is change in recorded placement episodes. Some of those changes involve a child moving and some do not. This report sets out how often the measure is met, which children meet it most often, what is recorded about why placements change, and what the national data cannot show.

8,500 of the 81,770 children looked after in England on 31 March 2025 had three or more placements during that year: 10% of children looked after on that date. In 2021 the figure was 7,400, or 9%. The denominator throughout is the 31 March snapshot, which is smaller than the number of children looked after at some point during the year.

The rate has stayed within roughly a percentage point of one in ten for five years, but the number of children has moved more: 1,100 more than in 2021, a rise of 14.9%, while the looked-after population grew 1.2%. The highest count was in 2024, at 8,860 children. The department cautions that 2021 shows the initial effects of the pandemic and that placement stability was slightly higher during it, so any comparison anchored on 2021 should be read with that in mind.

Within that group, 5,180 children had exactly three placements and 1,790 had four. 1,530 children, or 1.9% of the snapshot population, had five or more in twelve months, 410 of them seven or more.

A one-year measure captures less accumulated placement history than a two-year one. The department also publishes the measure over a two-year window, for the 59,430 children who were looked after on 31 March 2025 and had also spent at least one day in care in the previous year. That is not the same as having been in care for two years: a child in this cohort may have been looked after for a little over a year, and two years is the window their placements are counted over. Of them, 13,410 (22.6%) had three or more placements across that window, 1,270 of them seven or more. This is a different population from the one above, so the two rates are not one number growing.

The department splits the snapshot population by how long the child's current period of care had lasted. Among those looked after continuously for less than twelve months on 31 March, 14.1% had three or more placements: 3,280 of 23,240 children. Among those looked after continuously for twelve months or more it was 8.9%, or 5,220 of 58,530.

This is a current period of care, not a first-ever one. A child who had been looked after before, returned home and come back into care would appear in the shorter group, so these figures should not be read as describing children new to the care system.

The two groups also differ in how much time they had. A child whose current period of care began in January had three months in which to accumulate placements; a child looked after throughout had twelve. The group with less time reached the higher rate.

The two rates have moved in opposite directions. For the shorter group the rate has risen in each of the last three years, from 12.3% in 2022 to 14.1% in 2025. For the longer group it has fallen back from 9.6% in 2023 to 8.9%. The increase is concentrated in the group whose current period of care had lasted less than twelve months. These figures do not show why.

The department publishes two further measures of stability alongside placements: how often a child changes school, and how often they change social worker. Taking the same two-year cohort described above, the same window and the same threshold of three or more, 31,520 children (53%) had three or more social workers, against 13,410 (22.6%) who had three or more placements.

Over a single year the ordering is the same: 19,900 children, or 24.3% of all children in care, had three or more social workers, against 10.4% who had three or more placements. Among children looked after continuously for less than twelve months it reaches 32.8%.

School stability is measured differently and is not directly comparable: the threshold is one move rather than three, and it can only be measured for children matched to the school census. On that measure 5,910 of 52,130 school-aged children in care (11.3%) moved school at least once during the year, and 4,190 (8%) moved mid-year. The three kinds of instability overlap less than might be assumed: 310 children of 52,130 met the high-instability threshold on all three measures in a single year, and 1,660 of 45,680 (3.6%) over the two-year window.

Over a single year the differences between age groups are small. Over the two-year window the highest rate is among children aged 1 to 4, above the rate for those aged 16 and over. The two-year measure is not published for children under 1, who cannot have been in care across the whole window, so this is a comparison between the age groups that are published.

For very young children, a placement change may be part of a planned move towards adoption, a special guardianship order or a return home, and each of those steps is recorded as a placement. The measure cannot distinguish those changes from other kinds.

Legal status separates children more sharply than age does. Of the 2,160 children whose first legal status in the year was detained for child protection, 570 had three or more placements, or 26.4%. Among children on a care order the figure was 9.2% (5,320 of 58,100), and among children accommodated by voluntary agreement with their parents under section 20, 12.7% (2,440 of 19,150). These figures describe differences between legal-status groups as recorded in the year. They do not show why those differences arise, and the field is a child's first recorded legal status in the year rather than their route into care.

Because the number of unaccompanied asylum-seeking children in care has changed substantially since 2021, we tested whether that alone explained the national trend. It does not. Excluding them, the rate was 9.3% in 2021 and 10.8% in 2025, having peaked higher in between at 11.1% in 2023; those are endpoints, not the range. The comparable figures for all children are 9% and 10%. The rate for unaccompanied children is lower, 6.1% in 2025 (400 of 6,540), so as a matter of arithmetic their growing share has held the national rate down slightly. That is a compositional effect, not an account of any child's placement stability.

8.7% of children whose first placement of the year was a foster placement had three or more placements that year. For children whose year began in a children's home or secure home it was 13.6%, and for the small number whose year began in another residential setting, meaning a residential care home, an NHS setting, a mother and baby unit or custody, it was 24.3%.

This chart does not show that residential care is less stable than foster care, and it cannot. The categories describe where a child's year started, not where they lived, so these are different groups of children rather than the same children in different settings. The department also notes one exception to the definition: for up to 50 children a year it substitutes the placement on 31 March, mostly where the first placement of the year was respite or temporary.

The Arc analysis cannot establish why the groups differ. Previous linked-record research can speak to it: the study of care and education records by the Rees Centre and the University of Bristol found that very few of the young people who ended up in residential care had always been there, and that it was likely their placements reflected the use of that form of care as a consequence of fostering placements that had ended. Earlier placements ending may therefore be part of the pathway into residential care, which is a reason not to read the difference above as an effect of the setting itself.

The department publishes one measure that asks the opposite question: of children under 16 looked after continuously for at least two and a half years, how many had lived with the same carer for at least two years? In 2025, 19,680 of 28,710 children had, 69%. The remaining 9,030 children, 31.5%, had not met that continuity measure. It does not show whether their current placement was settled, suitable or intended to be permanent. The share was slightly lower in 2025 than in 2021, though it has not moved in one direction: it rose to 70.6% in 2022 before falling back.

Local authorities record a reason for every placement change. This is a separate measure from the one above: it counts all 59,140 recorded placement changes in the year, not the changes experienced by the 8,500 children who met the three-or-more threshold. The published data does not break the reasons down by that group.

Of those changes, 20,170 (34.1%) were recorded as a change to, or implementation of, the child's care plan, the largest single category. The code records what was entered on the return, not whether the change was planned in advance. The codes that describe a placement ending at someone's request are smaller. In the department's own category names: 5,010 (8.5%) were coded "carer requests placement ends due to child's behaviour", 1,870 (3.2%) "child requests placement end", 1,000 (1.7%) "responsible area/local authority requests placement end", 600 (1%) "standards of care concern", and 400 (0.7%) "resignation/closure of provision".

A further 14,570 recorded changes, 24.6%, were coded as "other", so the national table gives no more specific explanation for them. The department notes that this share has sat between 22% and 25% every year since 2021, and that placement changes due to custody have been reported under it. And 10,500 changes (17.8%) were the status-only changes described at the start of this report, in which the child stayed with the same carer.

How long placements lasted is a third measure again, counting placements that ended rather than children or changes. The figures below are for the year ending 31 March 2024 rather than 2025, because the newest year in any release is materially incomplete: reporting year 2024 gained 11,530 placements when it was revised, almost all of them long ones. The method section sets out the size of that effect. No trend in placement duration is published here.

A short placement is not necessarily a failure. Some are explicitly temporary: an emergency bed for a night, a bridging placement while an adoption is arranged, a few days before a child goes home. The categories that end soonest are ones where a short stay is often the point. "Other placement" had the highest share ending within a month, at 42.7%, and covers arrangements the department describes as court-directed placements with a relative, placements with unregistered carers, and unregulated placements made while a suitable one is found. Other residential settings were close behind on that threshold and had the highest share ending within a week; that category includes hospital and custody. Across all types, of the 110,410 placements that ended that year, 52.6% had lasted less than six months, 20.9% less than a month and 9.3% less than a week. 11.7% had lasted two years or more.

The Children Act 1989 requires a local authority, so far as is reasonably practicable, to find a placement that allows a child "to live near C's home", that "does not disrupt C's education or training", and that keeps siblings together. On 31 March 2025, 18,120 children, or 22.2% of all children in care, were living more than 20 miles from home.

Twenty miles is a blunt measure: equal distances can have very different practical consequences, and even a short move can disrupt school, friendships or family contact. The measure shows none of that. What it does show is a difference by placement type: 47.5% of the 9,480 children in children's homes and secure homes were more than 20 miles from home, against 19.5% of the 54,820 children in foster placements.

Where provision sits is part of the picture. Ofsted's Chief Inspector reported in 2025 that "more than a quarter of England's children's homes are in the North West, a region that accounts for just 18% of looked-after children". This data cannot show how far availability drives any individual placement decision.

Of the 153 local authorities looking after children in 2025, 150 had a publishable rate. They range from 4.7% to 17.9%, with a median of 10.4% and half of all authorities between 8.4% and 11.7%.

We have not published a table of best and worst councils, because the data will not carry one. Across the 144 authorities with a rate in every one of the five years, the median authority's rate moved 1.6 percentage points from one year to the next, and one in ten moved by 4.1 points or more, against a national rate near ten per cent. A single year's position is a weak signal.

Across all five years, 25 authorities were above the England rate in all five years and 30 were below it in all five. Even then, authorities differ in the age and needs of the children they look after, in how many unaccompanied children they are responsible for, in what placements exist near them and in how they code their returns, and none of that is visible in a rate.

Everything above is administrative data. It counts placements. It cannot say what a change felt like, what it cost a child, or whether it helped. The studies below speak to that. There are few of them, several are some years old, and none was designed to estimate the effect of a placement change; their findings are associations, reported as such by their authors.

An English study linking care records to educational records, carried out by the Rees Centre at Oxford and the University of Bristol using the 2013 GCSE cohort, found placement changes and school changes both associated with lower attainment, the association being strongest for changes during the exam years. Its authors were careful about what that means: "Both school changes and placement changes are risk factors for looked after children's educational outcomes... Both kinds of change may be markers of a child in difficulty." Placement change may be a marker of pre-existing difficulty rather than a cause of what follows. Nothing in the data published since resolves that, and this report does not claim otherwise.

On what children themselves report, the largest source is the Bright Spots survey run by Coram Voice and the Rees Centre, which collected 9,472 responses from children in care aged 4 to 18 across 38 English councils between 2016 and 2021. Among the young people aged 11 to 18, it found that always feeling settled was less likely for those who had experienced multiple placements: an association in a survey, not an estimated effect, and the councils and children taking part are not a random sample of children in care. Its open-text answers describe dimensions the administrative data omits. One child aged 8 to 10 wrote:

Three things in that account are not recorded nationally: how much notice a child was given, whether they met their new carers first, and what happened to their belongings. The same survey records children unsettled by "fears of having to move again, of important possessions that had been left behind and of missing their previous carers, their families and pets". A child aged 4 to 7 wrote: "I would like my old squeaky hippo, which I had when I was 5. It is at home with my mum, and I haven't seen it." The government's independent review of children's social care in 2022 published a care leaver's account of moving out of a foster placement with "bin bags to put ten years' worth of my belongings in".

The systematic review by Rock, Michelson, Thomson and Day, which covers foster placement instability rather than residential care, also identified factors associated with greater stability: placements with siblings, placements with older and more experienced foster carers with strong parenting skills, and placements where carers create opportunities for a child to develop intellectually.

Everything measured here is continuity. None of it measures whether a child is safe, cared for, listened to or happy, and a low placement-change rate is not evidence that any of those things are true.

The Bright Spots survey records children wanting opposite things. One young person asked "to change foster carers. Sometimes I don't feel like part of the family and feel like I don't belong with them." A child aged 8 to 10 wrote: "I want to be able to stay with my current foster carer, but my social worker says I have to move." Both appear in the national statistics only as a placement count, which cannot tell them apart. In the same survey, nine in ten of the children who said they never felt settled also said they did not feel safe where they lived.

That has a practical consequence. A system treating the three-or-more-placements measure as a target could improve it by keeping children in placements they have asked to leave. A system reading every change as a failure would discourage the changes that achieve permanence. The measure is useful for describing discontinuity and unsuited to being either.

Several important aspects of a placement change are not captured nationally at all.

Whether a child had notice. Whether they met their carers first. Whether their belongings arrived with them. Whether anyone asked them. Those are the parts of a change that children in the published research talk about, and the parts the national record is silent on.

Around one in ten children looked after in England on 31 March have met the three-or-more-placements threshold in each of the last five years, and the number of children behind that rate is higher than it was in 2021. Comparisons anchored on 2021 carry the department's pandemic caveat.

The rate is higher among children whose current period of care had lasted less than twelve months than among those looked after continuously for longer. The rate for the shorter group has risen in each of the last three years, while the rate for the longer group has fallen since 2023. These figures do not show why, and the shorter group is defined by the length of a current period of care rather than by being new to the care system.

The measure counts recorded placements, not necessarily changes of home. It cannot distinguish a change that harmed a child from one they wanted, one made as part of a plan for permanence, or one in which they stayed with the same carer. It should not be treated as a stand-alone measure of the quality of care.