Trends in the tribunal · 3
Out of time: how often tribunals let late claims in
What the published Employment Tribunal judgments show about extending time under the 'reasonably practicable' and 'just and equitable' tests: how often late claims get in, how long the delays were, and where representation made a difference.
Where the act complained of or the dismissal falls on or after 1 October 2026, most tribunal claims, including unfair dismissal, unlawful deductions and discrimination, now have six months instead of three (Employment Rights Act 2025; SI 2026/954). Earlier events keep three months, and every judgment studied here was decided under the three-month period. Either way, a late claim is not necessarily lost. The tribunal can extend time, but the test depends on the claim, and the two main tests are very different.
For unfair dismissal, unlawful deductions and most other claims under the Employment Rights Act 1996, the claimant must show it was not reasonably practicable to present the claim in time, and that it was then presented within a further reasonable period. The Court of Appeal has treated that as a question of fact for the tribunal on all the circumstances (Palmer and Saunders v Southend-on-Sea BC [1984] ICR 372), and it is applied strictly.
For discrimination claims, section 123 of the Equality Act 2010 allows whatever further period the tribunal thinks just and equitable. That is a broad discretion. Extending time is the exception rather than the rule (Robertson v Bexley Community Centre [2003] EWCA Civ 576), but there is no requirement for a good reason for the delay (Abertawe Bro Morgannwg UHB v Morgan [2018] EWCA Civ 640), and the familiar list of factors is a guide, not a checklist (Adedeji v University Hospitals Birmingham NHS Foundation Trust [2021] EWCA Civ 23).
Practitioners have a feel for how these tests play out. I wanted to see what the published judgments actually show.
The judgments
Starting from the full register of around 134,000 published judgments, I looked for every decision on whether to let a late claim proceed. About 9,700 contained both a time-limit test and a phrase suggesting lateness. A first reading by a local model found that 7,397 of those actually decided whether to let a late claim in, and picked out 10,773 separate decisions (one judgment can decide several). After the checks described below, 9,084 decisions were clearly either allowed or refused, and those are the ones the rates are based on. Each decision was read for the test applied, the outcome, the length of the delay, the reasons put forward and who represented the claimant.
Reading these is less straightforward than it sounds. On a first pass, about one in seven judgments (211 of 1,462) that seemed to contain an extension of time turned out not to: the claim had been in time all along, time was only discussed in the alternative, or the time point arose inside an application to amend rather than on a claim as presented. Those have been taken out. Uncertain decisions were re-read by a stronger model against the judgment text, and blind samples checked the rest. Of 200 refusals checked against the full judgment, 187 (93.5%) were confirmed as refusals, and in a sample of 50 confirmed decisions the stronger model agreed on 45 (90%). About 4,750 judgments whose only decisions were refusals were not re-read and were kept as refused, so the success rates may be slightly low, by up to about two points overall.
How often late claims get in
Under the reasonably practicable test, the late claim was allowed in about 9% of decisions (331 of 3,787). Under the just and equitable test, it was about 22% (1,149 of 5,260). Across all tests, about one late claim in six was allowed to proceed.
That gap is what the law would predict, but its size is worth noting. Under the strict test, a claimant who misses the deadline is very unlikely to recover from it. Under the broader discretion, roughly one in five succeeds.
The length of the delay
Where time was extended, the median delay was 54 days. Where it was refused, it was 91 days. The split holds under both tests: under the reasonably practicable test the medians were 34 days (allowed) and 60 days (refused), and under the just and equitable test 64 days and 169 days.
Shorter delays are more forgivable, which is unsurprising. More telling is how much longer the delays were under the just and equitable test: a successful discrimination claim was, on the median, later than an unsuccessful claim under the strict test.
What made the difference
Reasons were recorded for every allowed decision, but only for a minority of refusals, and those were not chosen at random, so the table covers successful extensions only. A decision can involve several reasons, so the shares add up to more than 100%.
| Reason accepted | Reasonably practicable: allowed decisions (of 331) | Just and equitable: allowed decisions (of 1,148) |
|---|---|---|
| Other reason | 136 (41.1%) | 341 (29.7%) |
| Ignorance of rights or time limits | 110 (33.2%) | 175 (15.2%) |
| Ill health | 88 (26.6%) | 317 (27.6%) |
| No reason recorded | 75 (22.7%) | 387 (33.7%) |
| Adviser’s fault or advice | 42 (12.7%) | 117 (10.2%) |
| Internal process or grievance | 19 (5.7%) | 246 (21.4%) |
| Awaiting another outcome | 9 (2.7%) | 16 (1.4%) |
| Adviser involvement | 1 (0.3%) | 0 |
One further just and equitable extension, found in a later check of the first-reading refusals, has no recorded reasons and is not counted here.
Ignorance of rights or time limits was the most common named reason under the strict test, behind about a third of successful extensions, while under the broader discretion internal processes and ill health weighed far more often (about one in five and one in four). A large share of successful decisions recorded no reason at all, most of them under the just and equitable test.
The examples below show each reason succeeding and failing. They come from the decisions I re-read in full.
Reasonably practicable
| Reason | Succeeded | Failed |
|---|---|---|
| Ill health | Baugh v IBM UK Ltd, para 48: “manifestly suffering a mental impairment” | Watt v Savills Management Resources Ltd, para 53: “undiagnosed ADHD and severe mental health deterioration” |
| Ignorance of rights or time limits | Cameron v Caremore (Aberdeen) Ltd, para 16: “genuinely ignorant of the existence of time limits” | Docherty v Royal Mail Group Ltd, para 87: “he could have checked the position with the union” |
| Adviser’s fault or advice | Quick v West Hertfordshire Hospitals NHS Trust, para 9.11: “incorrect advice about the first Acas certificate combined with technical difficulties” | Mitchell v Royal Mail Group Ltd, para 131: “She had access to trade union advice” |
| Internal process or grievance | Quinn v Asda Stores Ltd, para 9.4: “so lacking in awareness of his employment rights” | Fidalgo v London United Busways Ltd, para 27: “not sufficient excuse for the claimant to delay taking protective action” |
| Awaiting another outcome | Tasker v Duncan Adams Ltd, para 21: “he could not be expected, reasonably, to have understood that at the time” | Reynolds v Chief Constable of Devon and Cornwall, para 41: “ongoing internal appeal did not, in and of itself, render it not reasonably practicable” |
| No reason recorded | No clear example | Hallett v SP Planning Ltd, para 195: “no evidence to suggest that it would not have been reasonably practicable” |
Just and equitable
| Reason | Succeeded | Failed |
|---|---|---|
| Ill health | Conduit v Rosslyn Hill Unitarian Chapel, para 270.4: “absent sick due to stress” | Ekeruche v National Grid, para 243: “she was actively engaging” |
| Ignorance of rights or time limits | Rawat v Hare, para 13: “someone who was not aware of the action they were to take” | Sands v Mole Valley District Council, para 14: “In my judgment it was not reasonable” |
| Adviser’s fault or advice | Preston v EE Ltd, para 31: “The Claimant was lied to by Mr Kerslake via her trade union” | Ekeruche v National Grid, para 244: “had access to advisers” |
| Internal process or grievance | Cannon v South Lanarkshire Council, para 25: “understood from that advice that she required to undergo the grievance procedure” | Ojiako v H&M, para 92: “not a good reason for the delay” |
| Awaiting another outcome | Connolly v The Chancellor, Masters and Scholars of the University of Cambridge, para 127: “Given the Claimant did not get a final conclusion to the process” | Dave (Kacha) v CU Recruitment and Admissions Ltd, para 178: “she was awaiting a response from the Respondent, but that did not explain” |
| No reason recorded | Conduit v Rosslyn Hill Unitarian Chapel, para 270.4: “did not give direct evidence to explain the reason for this delay” | Khan v West London NHS Trust, para 126: “a conscious decision to put the matter behind him” |
Representation
The most interesting result concerns representation.
Under the reasonably practicable test, representation made almost no difference: claimants succeeded in about 9% of cases whether they had counsel, a solicitor or a lay representative, or represented themselves. That fits the test. Whether it was reasonably practicable to claim in time is a question about what happened before the deadline, and a representative arriving at the hearing cannot change those facts.
Under the just and equitable test, it was different. Claimants represented by counsel or a solicitor succeeded in roughly 30% and 28% of decisions, against 21% for claimants in person and 16% for those with lay representatives. Claimants who did not attend the hearing at all succeeded in about 2% of cases under the reasonably practicable test and 4% under the just and equitable test.
The data cannot say why. Represented claimants may have stronger cases, or cases a representative was willing to take on. But it is consistent with what a broad discretion would lead you to expect: when the tribunal is weighing prejudice, the merits and the reasons for delay, how well the argument is put may count for more.
The decisions can be searched in the extensions of time viewer, which charts success rates by claim year and filters by test, outcome, reason, delay and representation.