Judge Davis, Michael P.
EOIR data through August 31, 2026 · built September 11, 2026 · What these dates mean
Hearing cases: hearings are scheduled
- Last hearing
- August 27, 2026
- Next hearing
- September 2, 2026 · 14 scheduled in all
- Last decision
- December 30, 2025
Courts worked at: 2
Hearings scheduled here
- Sterling2022—2026 · 12,252 hearing records
Worked here before
- Annandale Immigration Court2022—2022 · 68 hearing records
The periods overlap, and that is not an error: a judge hears cases at a new court while still closing out old ones at the previous one. Each row answers "when did this judge work here", not "from when to when did they serve here".
Status is derived from the court calendar: whether the judge has hearings scheduled after the release cutoff. The EOIR release carries no personnel data, so retirement, a transfer and an extended absence cannot be told apart — only what the calendar shows is stated here.
How this is counted
Hearings are scheduled two to three years out, so an empty calendar for a working judge is rare; but for a judge who went quiet in the last few months it may simply be the data lagging. The lookup table's "active" flag is not used: it disagrees with the calendar for a third of judges.
35%
Asylum grant rate
of 403 decisions · FY2021–FY2026
95% interval: 31% — 40%
- granted:
- 142
- denied:
- 261
How this percentage is calculated
Grant rate = granted / (granted + denied). Administrative closures, withdrawals, abandonment and "not adjudicated" are excluded from the denominator. What is counted is a decision per proceeding, not per application: asylum, withholding and CAT protection are almost always asked for together, so withholding and CAT grants are counted separately from asylum. The U.S. fiscal year runs October 1 – September 30.
How this compares
- This judge35% (403)
- The whole court — Sterling24% (13,933)
- All U.S. courts33% (1,369,763)
This comparison does not adjust for which cases a judge was given: judges see different nationalities, different shares of detained respondents, and different rates of legal representation. A difference in percentages does not by itself mean identical cases were treated differently.
Trend by fiscal year
U.S. fiscal year: October 1 – September 30.
FY2026 is incomplete: the release covers only 11 of its 12 months. Its point is drawn hollow and the segment leading to it is dashed.
Show as a table
| Period | decisions | granted | 95% interval |
|---|---|---|---|
| FY2026partial | 26 | 4% | 1% — 19% |
| FY2025 | 112 | 18% | 12% — 26% |
| FY2024 | 152 | 47% | 39% — 55% |
| FY2023 | 100 | 45% | 36% — 55% |
| FY2022 | 13 | 38% | 18% — 64% |
By applicant nationality
| Nationality | decisions | granted | 95% interval |
|---|---|---|---|
| El Salvador | 117 | 30% | 22% — 39% |
| Honduras | 82 | 27% | 18% — 37% |
| Guatemala | 50 | 36% | 24% — 50% |
| Peru | 32 | 28% | 16% — 45% |
| Nicaragua | 21 | 38% | 21% — 59% |
| Brazil | 13 | 0% | 0% — 23% |
| Ghana | 13 | 69% | 42% — 87% |
| Mexico | 12 | 25% | 9% — 53% |
| Venezuela | 9 | 67% | 35% — 88% |
| Colombia | 9 | 56% | 27% — 81% |
| India | 5 | 0% | 0% — 43% |
| Azerbaijan | 4 | 100% | 51% — 100% |
| Nepal | 4 | 75% | 30% — 95% |
| Afghanistan | 4 | 100% | 51% — 100% |
| Haiti | 4 | 0% | 0% — 49% |
Timing and waits
How long each stage of a case takes before this judge. Every figure is a median: half of cases are faster, half slower. Data as of August 31, 2026; these are past and current cases, not a forecast for yours.
How long decided cases took
from the NTA to the decision
43 mohalf of cases: 23—65 modetailshidehalf of cases: 23—65 mo
How long decided cases took
from the NTA to the decision
- Cases decided
- 403
- Half of cases
- 23—65 mo
- The whole court
- 31 mo
| Decided in | cases | the case ran, median | half of cases, mo |
|---|---|---|---|
| FY2025 | 112 | 30 mo | 21—42 |
| FY2024 | 152 | 35 mo | 22—71 |
| FY2023 | 100 | 51 mo | 47—73 |
Cases decided this year may have begun a decade ago, so this is not a forecast for a case filed today.
The NTA is the charging document; serving it opens the case in court. The figure is a median, not a mean: durations have a long tail, and a few decade-long cases would pull a mean far from the typical case.
How long undecided cases have already waited
from the NTA to the data cutoff
too few to tellproceedings in the queue: 0nationwide 34 modetailshideproceedings in the queue: 0
How long undecided cases have already waited
from the NTA to the data cutoff
No undecided proceedings in this release.
The queue counts proceedings with no decision yet in the release. It is how long people have already waited, not how much longer they will.
Shown separately is the share with a hearing scheduled after the cutoff: the rest have no future event in the release at all, and cannot be read as live alongside these.
When the nearest hearing is set
from the data cutoff to the nearest hearing
too few to tellfewer than 50 casesnationwide 7 modetailshidefewer than 50 cases
When the nearest hearing is set
from the data cutoff to the nearest hearing
- Cases with a date set
- 9
- The whole court
- 3 mo
- Nationwide
- 7 mo · half: 2—15 mo
Fewer than 50 cases have a date set. There is nothing to base a median on: it would rest on two or three rows.
Hearings set shortly after the cutoff have already happened by the time you read this. The count here is of cases, while the queue above counts proceedings, so the two figures need not agree. The horizon describes the docket, not the speed of the court.
Rows attached to cases already decided are excluded — the release keeps them, and with them the count of those waiting would be a tenth higher. The judge and court come from the schedule row itself: whoever will hold the hearing, not whoever the proceeding is assigned to. Cases where the person is detained have a hearing within days, the rest within months, and courts at detention centres consist almost entirely of the former. The median is only given from fifty cases up: below that it rests on two or three rows.
From a master calendar hearing to the individual hearing
from a master calendar hearing to the one set after it
too few to tellfewer than 50 hearingsnationwide 11 modetailshidefewer than 50 hearings
From a master calendar hearing to the individual hearing
from a master calendar hearing to the one set after it
- Master calendar hearings in the year
- 133
- Followed by an individual hearing
- 15.8% (10.6% — 22.9%) · 21 of 133
- The whole court
- 9 mo
- Nationwide
- 11 mo · half: 2—24 mo
- Fewer than 50 hearings were followed by an individual hearing. There is nothing to base a median on: it would rest on two or three cases.
| What was scheduled next | hearings | share | median gap |
|---|---|---|---|
| An individual hearing | 21 | 15.8%10.6% — 22.9%14.7% nationwide | — |
| Another master calendar | 89 | 66.9%58.5% — 74.3%52.9% nationwide | 3 mo |
| Nothing: the proceeding is over | 23 | 17.3%11.8% — 24.6%27.5% nationwide | — |
| Nothing: no hearing set yet | 0 | 0.0%0.0% — 2.8%4.9% nationwide | — |
Under each share — the 95% interval.
The figures cover master calendar hearings over the last twelve months of the release and what was set after each of them. The shares describe hearings, not cases: one case passes through several master hearings in a year and is counted as many times.
The next hearing has usually not happened yet: its date can be moved. A separate line shows after how many hearings nothing is scheduled — either the proceeding is over or the calendar has not been filled in, and towards the end of the window the second share grows simply because the calendar has not caught up. Cases where the person is detained move more than twenty times faster than the rest, and courts at detention centres consist almost entirely of them. A slot counts the cases of one judge at one time: you go to your judge's courtroom, not to the building. No hearing dates are shown here in any form. The median is only given from fifty hearings up.
In absentia decisions
Removal cases completed without the respondent in the room. This is not about a judge's strictness: attendance depends on whether the person is detained, whether notice reached them and whether they have counsel.
34% of proceedings ended in absentia — 1,136 of 3,388.
- This judge34% (3,388)
- The whole court29% (80,591)
- All U.S. courts19% (11,588,297)
| Fiscal year | Proceedings | In absentia | Share |
|---|---|---|---|
| FY2021 | 13 | 0 | 0% |
| FY2022 | 80 | 6 | 8% |
| FY2023 | 764 | 215 | 28% |
| FY2024 | 1,749 | 611 | 35% |
| FY2025 | 709 | 285 | 40% |
| FY2026partial | 70 | 19 | 27% |
The denominator is every completed removal proceeding (RMV): other proceeding types almost never end in absentia. The headline share covers the whole history, while the table shows the latest fiscal years by the proceeding's completion date. An in absentia order can be challenged by a motion to reopen.
Bonds
The money that buys release from custody while the case goes on. An amount is not set in every decision, and the two figures — how often and how much — only read together.
1 bond decisions in September 2025 — August 2026; an amount was set in 0 of them. 100% are attached by a custody hearing.
An amount was set fewer than 25 times — 0 in all. There is nothing to state a median on: it would describe a handful of cases rather than the judge. The national median amount is $7,500.
How often an amount is set
- This judge, September 2025 — August 20260% (1)
- This judge, October 2021 — August 20250% (1)
- The whole court0% (1)
- All U.S. courts25% (91,185)
The code table covers September 2025 — August 2026; “was” in the amount column covers October 2021 — August 2025.
Under the share — the 95% interval; under the amount — the quartiles.
EOIR publishes no reference for the decision codes on this field — the release contains no lookup for it, so the codes are shown as codes, next to what has been measured. The amount does not depend on the judge alone: courts see different mixes of cases, criminal histories and circumstances of arrest.
How this is counted
"Now" and "before" stand side by side because over the past year both the median amount set and the share released from custody have moved markedly. What is measured under each code is how often an amount is set and how often the person was released from custody within sixty days of the decision: three codes carry an amount almost always and the rest almost never, and that gap is what the whole calculation rests on. The released share is counted not from every decision but from those whose sixty days have already passed: for a decision a week old "not released" would mean "we do not know". Payment itself is not recorded — "released" is read from the custody history, not from a record of a bond being posted. A median is stated from twenty-five amounts upwards, and the quartiles beside it show how widely the judge varies.
Full statisticsMotions, adjournment reasons, appeals to the BIA, three-member panels and second rounds after a remand.
Motions
What the judge does with requests made along the way: continue the hearing, allow appearance by video, terminate the case, put it back on the calendar. Judges differ more here than in the asylum grant rate.
2,921 motions decided in October 2021 — August 2026, 113 of them in the last twelve months. 75% are attached to the judge by the nearest hearing.
Thick line — This judge, thin line — nationwide.
The band is the Wilson interval: the share lies within it with 95% confidence.
Years below the 30 threshold: 2. They are not shown, and the line breaks there.
Show the numbers
| Fiscal year | Volume | Share | Nationwide |
|---|---|---|---|
| FY2025 | 66 | 73%61% — 82% | 71% |
| FY2024 | 123 | 84%76% — 89% | 80% |
| FY2023 | 178 | 86%80% — 90% | 79% |
| Motion | Decidednow / before | GrantedSeptember 2025 — August 2026 | BeforeOctober 2021 — August 2025 | Nationallynow / before |
|---|---|---|---|---|
| Motion to Dismiss | 2611 | — | 98%97% — 99% | 80%96% |
| Motion for Change of Venue | 8440 | — | 96%94% — 97% | 86%95% |
| Motion for Continuance | 15378 | — | 84%80% — 87% | 56%77% |
| Motion to Withdraw as Counsel | 12235 | — | 97%94% — 99% | 91%88% |
| Motion for Video Appearance | 11226 | — | 90%86% — 93% | 69%79% |
| Motion to Substitute Counsel | 9148 | — | 98%94% — 99% | 97%96% |
| Motion to Reopen | 8107 | — | 35%26% — 44% | 45%67% |
| Motion to accept late filed documents | 4110 | — | 98%94% — 99% | 86%90% |
| Motion for Administrative Closure | 13100 | — | 76%67% — 83% | 36%76% |
| Motion to Consolidate | 099 | — | 92%85% — 96% | 88%90% |
| Motion to Terminate | 792 | — | 65%55% — 74% | 59%71% |
| Motion to Recalendar | 571 | — | 86%76% — 92% | 85%88% |
Under each share — the 95% interval.
A dash means the window holds too few decisions to name a share; an empty recent column means the judge no longer decides motions of that kind. What the letters G and D mean, EOIR does not publish: there is no lookup for this field, and they are read from the behaviour of the cases.
How this is counted
The last twelve months and the years before them are never merged: motion practice changes faster than a case runs, and a single five-year number would average two different eras. A motion record carries no judge code of its own — it is taken from the nearest hearing by date, that is, from whoever was actually sitting on the case at the time; six motions in ten attach that way, the rest are credited to the judge of the proceeding, and the two methods disagree in one case out of five. The "nationally" column shows where judges genuinely differ: procedural motions are granted by nearly everyone nearly always.
Adjournment reasons
Why a hearing did not go ahead, and how far it pushed the case. The court assigns the reason itself; a hearing counts as adjourned when another one follows.
Of 752 hearings with a stated reason in September 2025 — August 2026, 606 were adjourned (81%).
Thick line — This judge, thin line — nationwide.
FY2026 is incomplete: 11 of twelve months are in the release, so its point is hollow.
Show the numbers
| Fiscal year | Volume | Median days, quartiles below | Nationwide |
|---|---|---|---|
| FY2026incomplete | 490 | 146 d67 d — 233 d | 93 d |
| FY2025 | 2,773 | 91 d70 d — 204 d | 161 d |
| FY2024 | 3,588 | 119 d78 d — 189 d | 146 d |
| FY2023 | 2,371 | 126 d83 d — 196 d | 132 d |
| FY2022 | 48 | 41 d8 d — 170 d | 120 d |
| Reason | Adjournmentsnow / before | ShareSeptember 2025 — August 2026 | Nationally | Follow-up | Median days |
|---|---|---|---|---|---|
| Respondent to seek representation | 41,664 | —was 19.2% | 9.3%15.3% | — | — |
| Docket management (postpone hearing) | 2981,295 | 49.2%45.2% — 53.1%was 14.9% | 9.7%8.3% | 84% | 179105–235was 92205 nationally |
| MC to IC - merits hearing | 341,031 | 5.6%4.0% — 7.7%was 11.9% | 12.0%10.2% | 100% | 151140–210was 259309 nationally |
| Vacated master calendar hearing | 13905 | —was 10.4% | 3.8%4.3% | — | — |
| Preparation - respondent/attorney/representative | 6741 | —was 8.6% | 9.1%8.1% | — | — |
| IJ leave | 20417 | —was 4.8% | 5.6%5.7% | — | — |
| IJ completion at hearing | 18378 | —was 4.4% | 8.3%4.5% | — | — |
| Preparation - DHS | 5369 | —was 4.3% | 1.6%1.2% | — | — |
| Other respondent/respondent's attorney/representative request | 11263 | —was 3.0% | 2.8%2.9% | — | — |
| Data Entry Error | 7204 | —was 2.4% | 4.0%4.3% | — | — |
| IJ reassignment | 15852 | 26.1%22.7% — 29.7%was 0.6% | 14.2%10.5% | 87% | 10543–237was 66134 nationally |
| Prosecutorial Discretion | 0196 | —was 2.3% | <0.1%1.0% | — | — |
The upper figure in a cell covers September 2025 — August 2026; the lower one and “was” cover October 2021 — August 2025.
Under the share — the 95% interval; under the median — the quartiles.
The reason does not assign blame: "respondent to seek representation" covers both a person looking for counsel and a court granting time to look. A dash means the window holds too few adjournments to name a share.
How this is counted
The recent window and the years before it are shown apart: adjournment practice changes faster than a case runs, and it is not only the frequencies that move but the reasons themselves, in different directions. Case-completion codes (8A, 8B, 8C) are separated from genuine continuances by behaviour rather than by wording: after them another hearing rarely follows, while after preparation continuances it almost always does. That is what the "follow-up" column shows, and a low value means "the case usually ended here", not "the adjournment failed". The median days figure is how far this reason pushed the next hearing; half the cases fall between the quartiles shown.
What became of appeals to the BIA
An immigration judge's decision can be appealed to the Board of Immigration Appeals. The key to this table is who filed: the same outcome means opposite things when the respondent appeals and when the government does.
| Filed by | Appeals | Decision stood | Sent back | Overturned | Closed without merits |
|---|---|---|---|---|---|
| Respondent | 35 | 91% | 0% | 0% | 9% |
| nationally | 712,355 | 74% | 13% | 1% | 11% |
There are also appeals of other kinds, not counted in this table: Appeal of IJ MTR — 8, Interlocutory Appeal — 2, Bond Appeal — 1, Bond MTR Reo — 1.
How often decisions are appealed
| Fiscal year | Granted | Appealed by the government | Denied | Appealed by the respondent |
|---|---|---|---|---|
| FY2022 | 5 | 0% (0.6%) | 8 | 75% (69.6%) |
| FY2023 | 45 | 0% (0.8%) | 55 | 76% (66.0%) |
| FY2024 | 71 | 4% (1.9%) | 81 | 74% (58.6%) |
| FY2025 | 20 | 5% (8.1%) | 92 | 89% (64.6%) |
| FY2026 | 1 | 0% (8.8%) | 25 | 92% (58.4%) |
In brackets, the same share across all courts. Counted by the date of the judge's decision rather than the Board's: an appeal is filed within a month, so recent years are nearly complete here — unlike the outcomes above.
Only appeals on the merits (Case Appeal) are counted: motions to the Board itself and bond appeals are different events. The same outcome means opposite things depending on the filer, which is why the bucket is called "the judge's decision stood", not "X won".
How this is counted
A dismissed appeal means the removal order stood when the respondent appealed, and that the asylum grant stood when the government did; the two do not go into one percentage. Recent years are undercounted: an appeal takes months or years to resolve, so outcomes are counted by the date of the Board's decision while the appealed share is counted by the date of the judge's decision.
Board panels
By default a single Board member decides an appeal. Referral to a three-member panel marks a case found difficult: not how the appeal ended, but how it was heard.
Appeals on the merits decided in September 2025 — August 2026: 27.
Fewer than 50 appeals decided — 27 in all. There is nothing to state a share from: referral is a rare event, and on so few the interval is wider than the whole spread between judges. Nationally 4.1% of appeals are referred.
What follows shows in the outcomes: of 9,114 referred appeals the case went back to the judge in 60.5%, against 10.1% for the rest. The release carries no grounds for referral, so what stands here is measured behaviour, not a decoded label. This is an observation, not an explanation: a panel is not stricter, it gets different cases.
How often an appeal goes to a panel
- The whole court0% (467)
- All U.S. courts4% (31,254)
By who filed the appeal
| Filed by | Appealsnow / before | ReferredSeptember 2025 — August 2026 | BeforeOctober 2021 — August 2025 | Nationally |
|---|---|---|---|---|
| Respondent | 278 | — | — | 2.6%7.2% |
Under each share — the 95% interval.
The grounds for referral are not in the release: the table has no column pointing at a lookup, so in place of a reason stands the measured behaviour of the cases. And this is not a property of the judge alone: panels take cases of a particular kind, and judges see different mixes of them.
How this is counted
The grounds themselves are listed in 8 CFR 1003.1(e)(6) — all of them about complexity or a departure from law. Counting goes by the date of the Board's decision, not the date the appeal was filed: referral happens between the two, and an unresolved appeal may lack the marker simply because it has not been reached yet. The two windows are shown apart because the referral rate has moved markedly over the past year, and almost entirely among respondent appeals: without the filer axis the shift would read as a general one. A share is stated from fifty decided appeals in a window upwards: the event is rare, and on fewer the interval is wider than the whole spread between judges.
How to cite this page
Figures on this page belong to one monthly release and are replaced whole by the next. The release label and the data cutoff are part of the citation for that reason: without them the number cannot be checked later.
Asylum Beacon. Judge Davis, Michael P. EOIR Case Data, U.S. DOJ FOIA Library, release 2026-08, data through 2026-08-31. https://asylumbeacon.com/en/judge/MPD
Data comes from the monthly EOIR Case Data release (U.S. DOJ FOIA Library). This service is not affiliated with the U.S. government and does not give legal advice. The release lags by roughly a month, so check ACIS or Respondent Access for your current case status. The source has known gaps, including among relief applications: if something is missing here, that does not mean it is missing from your case.