How these figures are counted
Every number on these pages is counted from the case data the Executive Office for Immigration Review publishes each month in its FOIA library. The pages now use the release of October 1, 2026 and cover proceedings completed from 2022 through 2026 (October 2026 release). They are rebuilt when EOIR publishes a new release.
Which judges have a page
A judge has a page with at least 30 completed proceedings in the period and a personal name in EOIR's list of judge codes. EOIR also records proceedings under codes that aren't one person, such as visiting-judge codes; those count in a court's and the country's figures and have no page. A court has a page with at least 30 completed proceedings in the period. A judge is listed under the one court where the judge completed more proceedings in the period than at any other, when that court has a page. That is a count, not a statement of the judge's present assignment, and these pages don't say who sits where today.
A page's address is EOIR's code for the judge or court, so it stays the same when a name is corrected. A judge with no completion in the newest twelve months of the data is marked. The data doesn't say whether a judge retired, was reassigned or received a new code.
Asylum decisions
An asylum decision is a completed proceeding in which the judge granted or denied an application for asylum, withholding of removal or protection under the Convention Against Torture. A proceeding counts once, however many of those applications were in it:
- Asylum granted when asylum was granted.
- Withholding or Convention Against Torture protection granted when asylum was not granted and one of those was.
- Denied when none was granted and at least one was denied.
The share of decisions in which asylum was granted is given for two periods: the newest twelve months of the data, now October 2025 through September 2026, and the whole period, 2022 through 2026 (October 2026 release). Grant rates change from year to year, so a rate over the whole period depends on the years in which a judge decided cases, and the twelve months show the judge's recent decisions. The twelve months end with the newest month that has a completion before the release, and their decisions are counted month by month from the same proceedings as the yearly figures. A judge or court with no asylum decision in those months is marked "No decisions."
Three kinds of proceeding are left out of the rates and counted together on their own line: those where the applications were withdrawn, abandoned or otherwise ended with no grant or denial recorded; those recorded with a decision code we can't classify; and those recorded with both a grant and a denial of asylum. That line doesn't mean the judge made no decision in them.
The average time in court is an estimate for each proceeding, not for the whole case: it runs from the date the proceeding was entered in the court's records (or, when that is missing, the date of the charging document or of a hearing) to the proceeding's completion. It covers every completed proceeding with one of these applications that has a usable start date, decided on the merits or not. A case reopened or sent back after an appeal starts a new proceeding with a new clock.
How removal proceedings ended
Every completed removal proceeding ends with a judge's decision or with a completion that isn't one. The endings use EOIR's own codes for removal cases, from the lookup table it publishes with the data (also in its code key): removal, removal with withholding of removal or Convention Against Torture protection granted, voluntary departure, relief granted, terminated, dismissed by the judge, and a few others. A removal order is counted as in absentia when EOIR's data marks that the person did not appear; an order without that mark is not proof that a hearing was held.
The unit is a proceeding, not a case: a case that is sent back and decided again counts each time. The percentages are of the proceedings a judge decided. Changes of venue and transfers, cases EOIR completes as "failure to prosecute (DHS cases only)", and administrative closures are counted on their own lines. Other kinds of case, such as credible-fear and reasonable-fear reviews and withholding-only proceedings, use the same codes with different meanings, so they are counted and not broken down.
Other relief
Cancellation of removal, adjustment of status, waivers, suspension of deportation with NACARA relief, and voluntary departure are counted the same way: a proceeding counts once for each kind of relief the judge granted or denied in it.
Appeals
Appeals are counted with the proceedings a judge completed in the period, whichever side appealed, and by what the Board of Immigration Appeals has done with them so far. EOIR's data ties some appeals to a case and not to one proceeding; those are counted with the newest proceeding in the case, so a few may concern another judge's decision. A remand sends a case back to the immigration court for more proceedings. It isn't a finding of error and shouldn't be read as a reversal rate. Dismissals and denials include those with a grant of voluntary departure. Appeals from the newest years may still be pending, so their outcomes will change.
Minimum counts
A percentage is printed only over at least 30 decisions. Below that, the page gives the counts alone. A percentage that would round to 0% or 100% while some decisions went the other way is printed as "under 1%" or "over 99%."
Reading a judge's rates
Rates differ between judges for reasons that have nothing to do with the judge. Detained dockets, dockets of people without a representative, and dockets with different nationalities produce different outcomes, and judges are assigned different dockets. Each judge's page therefore shows how many of the judge's decisions involved people who were detained, who had a representative's appearance on the case record, and where they were from, and gives the court's figure for the same group. The representative's appearance is EOIR's case-level record (form EOIR-28): it shows that a representative appeared in the case, not that one was present when the judge decided. Nothing here predicts how a judge will decide a case.
Backgrounds
Where EOIR announced a judge's appointment in a notice with biographical information, the judge's page quotes EOIR's paragraph, unchanged, with the notice's date and a link to it. EOIR's news page links the notices; the pages now draw on those from May 2024 and from October 2025 on. A judge is matched to a notice by name, and nothing is shown when the match isn't certain: when two judges could answer to the name, when EOIR's data now records the judge under a different name, or when the judge's own record doesn't fit the notice (a judge announced to begin in one year who had a full docket in an earlier one). Judges appointed earlier, and judges whose names EOIR's data records differently, have no background here. We don't write or summarize biographies ourselves.
What these pages leave out
The case data has no decisions' text, no reasons, and nothing about a judge's background. It is administrative data keyed by court staff, and it contains errors. A judge's page can show some proceedings completed before the judge took the bench, mostly changes of venue and transfers: EOIR's data appears to credit the judge now assigned to a case with some of that case's earlier proceedings. Everything is counted as recorded, in the year recorded for it, with one exception: a completion dated in a year after the release isn't counted. A judge's newest completion is given by month, from EOIR's monthly counts, and a month after the release is never used.
Corrections
If something here is wrong, write to chris@passimai.com with the page's address. These pages are a research aid, not legal advice.