What changed in this release
The court publishes no change log — it republishes its entire database each month. We keep the previous copy and compare the two, which shows what moved during the month: where the judge changed, where a hearing date appeared, where a decision was entered.
This compares two monthly releases; it is not a log of what happened in court. EOIR publishes no change log — it republishes the whole database each month, so a previous value simply disappears and nothing records that it changed. The difference is visible only to someone keeping both copies. Hence two caveats. An event is dated by the release, not by the day it happened in court: roughly three weeks pass between the data cutoff and the publication of the dump. And each row states an observation only — "the proceeding has a different judge", not "the case was reassigned"; the release carries no reasons.
What happened during the month
Counted over cases and proceedings present in both copies. Each row is an observation, not a cause: the release says what became different, not why.
| Event | Count | Cases |
|---|---|---|
| Proceeding has a different judge | 219,857 | 217,185 |
| First hearing scheduled | 208,579 | — |
| Motion decided | 149,071 | 121,149 |
| Decision entered on a proceeding | 132,594 | — |
| Next hearing date moved | 93,918 | — |
| New cases in the release | 54,813 | — |
| Proceeding has a different court | 37,480 | 36,261 |
| New proceeding on an existing case | 34,251 | 32,602 |
| Appeal filed with the BIA | 16,986 | 16,946 |
| Bond decided | 8,804 | 8,167 |
| BIA appeal decided | 3,687 | 3,668 |
| Scheduled hearing no longer listed | 697 | — |
| Case no longer in the release | 556 | — |
| Proceeding no longer in the release | 714 | 713 |
| Decision removed from a proceeding | 160 | — |
Which way dates move
Among cases whose next hearing was scheduled in both the previous copy and this one.
- To an earlier date93,462 (99.5%)median 209 days, half the cases within 65–432 days
- To a later date456 (0.5%)median 66 days, half the cases within 28–125 days
Movement is almost always toward an earlier date, and usually it is not a rescheduling. A new, earlier hearing appeared on the calendar and became the next one for that case. The release shows the result, not the court's action — which is why this says "the date moved", not "the hearing was rescheduled".
The backlog
Proceedings with no completion date — what the courts still have to hear.
- In the previous release
- 3,195,566
- In this one
- 3,141,820
- Decision entered
- 132,594
- Newly opened, still undecided
- 79,397
- Decision removed
- 160
- Disappeared while undecided
- 709
Of 132,594 new decisions, 6,826 (5.1%) are dated on or before the previous cutoff: the previous release should have shown them and did not. Another 11,921 proceedings arrived already decided — they never spent a day in the backlog.
How much the source rewrites the past
Over months both releases cover in full. A plus means the new copy holds more records for that month than the previous one did.
| Month | Asylum decisions | Completed proceedings | Hearings | Motions decided |
|---|---|---|---|---|
| Jun 2026 | +6.5% | +4.3% | 0 | +6.5% |
| May 2026 | +1.6% | +0.2% | 0 | +0.4% |
| Apr 2026 | +1.0% | +0.1% | 0 | +0.2% |
| Mar 2026 | +0.8% | 0 | 0 | +0.1% |
| Feb 2026 | +0.9% | 0 | 0 | +0.1% |
| Jan 2026 | +0.8% | 0 | 0 | 0 |
| Dec 2025 | +0.5% | 0 | 0 | 0 |
| Nov 2025 | +0.5% | 0 | 0 | +0.1% |
| Oct 2025 | +0.6% | 0 | 0 | +0.1% |
| Sep 2025 | +0.6% | 0 | 0 | 0 |
| Aug 2025 | +0.6% | 0 | 0 | 0 |
| Jul 2025 | +0.7% | 0 | 0 | 0 |
| Jun 2025 | +0.9% | 0 | 0 | 0 |
| May 2025 | +0.8% | 0 | 0 | 0 |
| Apr 2025 | +0.7% | 0 | 0 | 0 |
| Mar 2025 | +0.5% | 0 | 0 | 0 |
| Feb 2025 | +0.9% | 0 | 0 | 0 |
| Jan 2025 | +1.0% | 0 | 0 | 0 |
| Dec 2024 | +1.5% | 0 | 0 | 0 |
| Nov 2024 | +0.9% | 0 | 0 | 0 |
| Oct 2024 | +0.7% | 0 | 0 | 0 |
| Sep 2024 | +0.8% | 0 | 0 | 0 |
| Aug 2024 | +0.9% | 0 | 0 | 0 |
| Jul 2024 | +0.9% | 0 | 0 | 0 |
How much the source rewrites what it already published. Measured over months both releases cover in full: if the new copy reports more decisions for March than the previous one did, the difference is records entered after the fact. The measures behave differently, and that is the answer: the calendar is barely revised, while decisions are revised more the closer the month is to the cutoff. The practical consequence — the most recent month is always incomplete and cannot be compared with earlier ones.
Across releases
A single release is a snapshot and cannot say whether the backlog is growing or shrinking. The series starts from the month we began keeping copies.
| Release | In the backlog | Of those, unscheduled | Hearings scheduled ahead | Asylum grant rate |
|---|---|---|---|---|
| 2026-06 | 3,195,566 | 426,712 | 3,799,951 | 33.8% |
| 2026-07 | 3,141,820−53,746 | 423,645 | 3,791,019 | 33.4% |
The series holds 2 points; a direction is worth drawing from 3. For now this is the difference between two snapshots, not a trend.
A series across releases. A single release cannot answer whether the backlog is growing or shrinking — it is a snapshot. Only a series can, and it starts from the month we began keeping both copies, not from the start of the data. Until there are three points no direction is drawn: the difference between two snapshots is one point, not a trend.
Data is current as of July 31, 2026