EOIR Insights

Immigration Judge Carmen Rey Caldas

Lumpkin, GA Immigration Court · Former judge

About this page
What is EOIR? The Executive Office for Immigration Review — the Justice Department agency that runs the nation's immigration courts and their appellate body, the Board of Immigration Appeals. More about EOIR
Read this first. These are past decisions, not predictions, and not a tool for choosing or avoiding a judge. How to read these figures

Immigration Judge Carmen Rey Caldas

Lumpkin, GA Immigration Court · Former judge Fired · August 2025
Appointed March 2022 · President Joe Biden → Attorney General Merrick Garland
4 years on the immigration bench
Appointed under a Democratic administration

No longer on the bench

Judge Carmen Rey Caldas was fired in August 2025, as reported by Gothamist (September 26, 2025).

In EOIR's case records, the last month in which this judge decided cases at their usual volume was August 2025 (records run through September 2026).

One of 60 immigration judges reported fired or pushed out since January 2025. The statistics on this page describe the judge's decisions while on the bench.

Check your hearing date

The court's own system has the current date, time and place of your next hearing. Hearings are rescheduled without much notice; check again the week before.

Look up my hearing (A-number) ↗
Call 1-800-898-7180Automated, in English and SpanishHave your A-number ready
Missing a hearing can end in a removal order entered in your absence. If you cannot attend, your lawyer can ask the court to move the date — before the hearing, not after.

Where the court is

Lumpkin Immigration Court
146 Cca Road, Po Box 248
Lumpkin, GA 31815

Court phone: (229) 838-1324

Open in Maps ↗
Arrive early: building security and the courtroom check-in take time. Bring photo ID and every notice the court has sent you.

Background

The Honorable Immigration Judge Carmen Rey Caldas was appointed as an immigration judge in March 2022 by Attorney General Merrick B. Garland. She earned a Bachelor of Arts in 2001 from New York University and a Juris Doctor in 2006 from Brooklyn Law School. She began her career at Sanctuary for Families in New York, serving as an Equal Justice Fellow (2006-2008) and staff attorney (2008-2011), and later as a senior immigration attorney at Her Justice (2011-2015). She directed the Immigration Intervention Project at Sanctuary for Families (2015-2018), co-directed the Safe Harbor Clinic as an assistant professor of clinical law at Brooklyn Law School (2018-2020), and served as U.S. Legal Director at the International Refugee Assistance Project (2020-2022). She is a member of the New York State Bar.

This judge in the news

Articles about this judge, and things the judge has written. Each link goes to the original source.
In the news
Official announcement
  • EOIR Announces 25 New Immigration Judges U.S. Department of Justice, EOIR · March 25, 2022

    EOIR announced Carmen Maria Rey Caldas's appointment to the Stewart Immigration Court, citing roles at the International Refugee Assistance Project, Brooklyn Law School and Sanctuary for Families.

Search reviewed October 16, 2026.

Caseload & process (since 2021)
1,063
Completions per year
5,398 total over 5.1 years
43%
Detained at decision
share of cases
2.1 mo
Median time to decision
from case start
6.7%
Ordered removed in absentia
person absent
2.2%
Decisions appealed to BIA
appeal rate
7.4%
Respondents’ appeals sent back or reversed
2 of 27 appeals decided
60%
DHS appeals sent back or reversed
12 of 20 appeals decided
2,068
Cases pending now
assigned to this judge, undecided
8 months
Typical wait for next hearing
half the open cases have their next hearing sooner, counting from October 1, 2026, when EOIR's data ends
3.1 years
Furthest hearing set
the latest next hearing on the calendar: October 29, 2029
1
Hearings per case
median per decided case, master and individual hearings before this judge; 1 in 4 took 2 or more; 3 when the case reached an individual (merits) hearing; 167 of the 1,793 cases had no hearing before this judge on record and count as 0

How this judge's cases ended

Every removal-type proceeding this judge completed since 2021, by how it closed. "Moved to another court" is a change of venue or transfer, not a decision.
Ordered removed
1,108 (25%)
Relief granted
244 (5.5%)
Terminated or dismissed
392 (8.8%)
Voluntary departure
94 (2.1%)
Moved to another court
2,318 (52%)
Administratively closed
63 (1.4%)
31% of the removal orders this judge entered (344 of 1,108) were entered in absentia, with the person not at the hearing. If you cannot attend a hearing, tell the court and your lawyer before the date, not after.
4,462 proceedings. Relief includes asylum, withholding, cancellation, adjustment and other grants.

Due process indicators

Since January 20, 2025. What this judge has done with the requests that decide whether a person gets a fair hearing. Every figure is about this judge alone; none is ranked against colleagues.
What the records show

None of the 10 warning signs appears in this judge’s logged rulings; 1 could not be checked.

  • Continuance requests from respondents’ lawyers
    Measures whether a person with a lawyer gets time to prepare, gather evidence and pursue relief.
    92% granted (94 of 102). Warning sign if under 33% (fewer than 1 in 3).
    Source: EOIR’s motions table. Motions for a continuance (type CO) filed by the respondent or their representative, with a decision of granted or denied, received since January 20, 2025. Only motions filed while a lawyer was already on the case count (an E-28 notice of appearance dated on or before the motion): a request filed without a lawyer is usually the routine first reset to find one. Motions withdrawn or found moot are left out, and oral requests decided in court are often never logged, so this counts what EOIR recorded, not every request made. Warning sign when fewer than 33% were granted. Under 20 rulings the row carries the low-sample note but is still judged.
  • Continuances before and since January 20, 2025
    Measures whether the judge gives people less time since January 20, 2025 than before.
    88% before, 92% since: up 4 points. Warning sign if down 20 points or more.
    The same continuance motions, with the judge’s grant rate from January 1, 2021 up to January 20, 2025 compared with the rate since. Warning sign when the rate since is 20 points or more below the earlier rate, on the whole-number percentages shown. Judges with no rulings before January 20, 2025, mostly those appointed after it, get a dash. The reason for a drop, a change in the docket or pressure from above, is not in the records; the drop is noted regardless.
  • Same request, different answer: DHS against the respondent’s lawyer
    Measures whether the judge says yes to the government and no to the person on the same kind of request: more time, a late document, an extension, appearing by video or phone, an earlier hearing.
    DHS 100%, respondent 91%: DHS 9 points higher. Warning sign if DHS is 25 points or more higher.
    Low sample: 8 rulings
    The kinds of motion both sides file: continuance (CO), accept a late-filed document (MD), extension of time (ET), video (TV) or telephone (TA) appearance, and advance the hearing (EH), since January 20, 2025, granted or denied. DHS’s motions of these kinds are compared with the same kinds filed by the respondent’s lawyer (a lawyer on the case when filed). Change of venue, dismissal, termination and reopening are not part of this comparison. Warning sign when DHS’s grant rate is 25 points or more above the respondent’s. DHS files far fewer of these, so DHS’s side often rests on a few dozen rulings; under 20 on either side the row carries the low-sample note.
  • Respondents’ motions to reopen, reconsider, administratively close, stay removal, suppress evidence or go on the status docket
    Measures whether the judge gives the person’s own motions a real chance.
    94% granted (48 of 51). Warning sign if under 25% (fewer than 1 in 4).
    The respondent’s motions to reopen (O), reconsider (FD), administratively close (AC), stay removal (SY), suppress evidence (SP) and place the case on the status docket (PS), since January 20, 2025, granted or denied, whether filed by a lawyer or by the person. Motions to terminate and for voluntary departure are not included. Warning sign when under 25% were granted. Motions to reopen are most of this pile and are denied more often everywhere, so a judge with many of them, for instance after in-absentia orders, will show a lower rate.
  • DHS motions to pretermit asylum (end it without a full hearing)
    Measures how often the judge lets DHS end an asylum claim without a full hearing.
    No rulings on record. Warning sign if 90% or more.
    DHS motions to pretermit (PM), which ask the judge to end an asylum or withholding claim without a full hearing on the merits, since January 20, 2025, granted or denied. Warning sign when 90% or more were granted. A judge who pretermits on their own motion, without a DHS filing, does not appear here. Pretermission motions were rare before 2025, so there is no earlier rate to compare.
  • Asylum decided on the merits since January 20, 2025
    Measures whether asylum is a real possibility before this judge.
    64% granted (85 of 133). Before: 47% (74 of 156). Warning sign if under 10%.
    Asylum applications this judge decided on the merits since January 20, 2025: granted against denied, one decision per proceeding, dated by the proceeding’s completion; the same definition as the asylum rate at the top of the page, limited to this window. Withdrawn, abandoned and otherwise closed applications are not counted, and neither are grants of withholding or protection under the Convention Against Torture. Warning sign when under 10% were granted. The judge’s rate from January 1, 2021 up to January 20, 2025 is shown beside it for context and does not change the mark.
  • Asylum denials within a year of the Notice to Appear
    Measures how often the judge rushes to deny an asylum case.
    8% of denials (2 of 24) came within a year. Warning sign if 25% or more. Detained cases are left out.
    Asylum denials since January 20, 2025 in proceedings where the person was not detained, one per proceeding, and how many came within 365 days of the Notice to Appear (the charging document’s date to the proceeding’s completion). Detained cases are left out because those dockets move fast by design. Warning sign when 25% or more of the denials came within a year. EOIR’s own expedited dockets can raise a judge’s share, and the records do not show whether the person asked for more time first.
  • Typical bond amount
    Measures whether this judge sets bonds far above what judges set elsewhere.
    Median $3,000 across 15 bonds set; the national median is $7,000. Warning sign if $15,000 or more (double the national median).
    Low sample: 15 bonds set
    Bonds this judge set since January 20, 2025 (EOIR’s bonds table: decisions that set or adjust an amount, the amount as recorded, the same rows as the “Bond in depth” section). Warning sign when the median is $15,000 or more, double the national median since that date ($7,000). Only the median is published, never the average. Under 20 bonds the row carries the low-sample note.
  • Bonds of twenty-five thousand dollars or more
    Measures how often this judge sets bonds that almost no one can pay.
    0% of bonds set (0 of 15). Warning sign if 25% or more (1 in 4).
    Low sample: 15 bonds set
    The same bonds set since January 20, 2025, and how many were $25,000 or more. Warning sign when those are 25% or more. Nationally about 4% of bonds set since that date reach that amount.
  • Bond granted when decided on the merits
    Measures whether a detained person has a real chance of release on bond before this judge.
    85% granted (17 of 20). Warning sign if under 20% (fewer than 1 in 5). Requests turned away for lack of jurisdiction are not counted.
    Bond requests this judge decided on the merits since January 20, 2025: granted (amount lowered, release on recognizance, amelioration granted) against denied (no change, no bond, amelioration denied). Moot, withdrawn and no-jurisdiction rulings are left out: since Matter of Yajure Hurtado (BIA, September 2025) judges turn away most requests from people who entered without inspection for lack of jurisdiction, and that follows the Board, not the judge. Warning sign when under 20% were granted. Who still reaches a merits decision changed with that ruling, so this sign is less clean than the two above it.
Continuance requests from respondents’ lawyersFiled with a lawyer already on the case: time to prepare, gather evidence or pursue relief. Requests to find a lawyer are not counted.Usually grants
92% 94 of 102 granted
Before January 20, 2025: granted 90 of 102 (88%).
Motions from each side
Respondent’s motionsUsually grants
94% 1,022 of 1,092 granted
DHS’s motionsUsually grants
96% 310 of 322 granted
Respondents’ motions to terminate the case
83% 85 of 102 granted
DHS motions that end a case quickly
Change of venue (transfer to another court)
99% 90 of 91 granted
Dismiss the case at DHS’s requestRespondents often want this too; since May 2025 DHS also uses it to move people into expedited removal. The records don’t say which.
86% 12 of 14 grantedLow sample: 14 rulings
13 of these were people with no lawyer on record; 11 granted.
Pretermit asylum (end it without a full hearing)
No rulings on record
Respondents’ appeals to the Board (BIA)Appeals of this judge’s decisions. Motions to reopen filed with the Board itself, bond and interlocutory appeals, and remands only for background checks are not counted.
0%
The Board sent back or reversed 0 of 4 decided appeals (0%).Low sample: 4 appeals decided Before the cut-off: 2 of 24 (8%).

How this is counted: motions EOIR logged with a decision of granted or denied (moot and withdrawn motions are left out). Many oral motions are decided on the record without an entry, so these figures count what was logged, not every request made in court. Every rate and label is shown however few rulings it rests on; under 20 rulings it carries a “Low sample” note. Labels: never (none granted), rarely (under 25%), sometimes (under 75%), usually (75% or more), always (every one granted). EOIR data through October 1, 2026. Why these figures matter: No Due Process in Immigration Court System.

55%
of asylum claims granted on the merits
Based on 289 asylum decisions (grants + denials) since 2021.
This judge
55%
This court
16.9%
National average
28.1%
0%100% granted
In the top 12% of the 961 judges nationally with enough cases to compare.
Grants asylum 1st most often of 8 judges at this court with enough cases to compare.

Does a lawyer matter here?

Asylum grant rate by whether the person had a lawyer on file when the case was decided.
With a lawyerrepresented · 221 decisions since 2021
65.2%
Without a lawyerunrepresented · 68 decisions since 2021
22.1%

Asylum grant rate over time

Each point is that calendar year's grant rate, back to 2015 where the judge was deciding cases — the longer career view. Dashed grey line = current national average (28.1%). Headline rates above use 2021 – today.
0% 25% 50% 75% 100%2016: 50% of 2 decisions · Low sample: 2 decisions50%20162017: 0% of 1 decisions · Low sample: 1 decision0%20172018: 0% of 1 decisions · Low sample: 1 decision0%20182019: 0% of 5 decisions · Low sample: 5 decisions0%20192020: 0% of 1 decisions · Low sample: 1 decision0%20202022: 40.5% of 37 decisions · Low sample: 37 decisions40.5%20222023: 35.3% of 34 decisions · Low sample: 34 decisions35.3%20232024: 53.2% of 77 decisions53.2%20242025: 67.9% of 134 decisions67.9%20252026 (partial year): 0% of 7 decisions · Low sample: 7 decisions0%2026*
* 2026 is a partial year (data through 10/01/2026) and is incomplete.

Asylum outcomes by nationality

Of the asylum applications this judge decided on the merits, the share granted, by the applicant's nationality. For people from China, 80% means this judge granted asylum in 36 of 45 decisions. Only grants and denials count: withdrawn, abandoned and undecided applications are left out, and a person denied asylum but granted withholding of removal or protection under the Convention Against Torture counts as a denial here. Every rate is shown; one resting on fewer than 20 decisions carries a “Low sample” note. "All judges" is the same measure for every immigration judge in the country, with all their decisions for that nationality added together.
China45 decisions
80%
All judges 63.5%
EcuadorLow sample: 19 decisions
47.4%
All judges 12.4%
IndiaLow sample: 19 decisions
73.7%
All judges 46.7%
HondurasLow sample: 17 decisions
11.8%
All judges 16.1%
El SalvadorLow sample: 15 decisions
13.3%
All judges 24.7%
NicaraguaLow sample: 14 decisions
64.3%
All judges 27.4%
RussiaLow sample: 14 decisions
92.9%
All judges 70%
MexicoLow sample: 12 decisions
0%
All judges 8.2%
Show all 46 nationalities
BangladeshLow sample: 10 decisions
80%
All judges 57.5%
ColombiaLow sample: 10 decisions
80%
All judges 11.9%
AfghanistanLow sample: 9 decisions
66.7%
All judges 70%
VenezuelaLow sample: 9 decisions
33.3%
All judges 30.2%
GeorgiaLow sample: 8 decisions
87.5%
All judges 47%
GuatemalaLow sample: 8 decisions
0%
All judges 19.6%
SyriaLow sample: 7 decisions
85.7%
All judges 56.7%
JamaicaLow sample: 6 decisions
16.7%
All judges 18.6%
PakistanLow sample: 6 decisions
100%
All judges 51.3%
SenegalLow sample: 6 decisions
50%
All judges 11.3%
UzbekistanLow sample: 6 decisions
66.7%
All judges 39.3%
GuineaLow sample: 5 decisions
60%
All judges 34.7%
CubaLow sample: 4 decisions
0%
All judges 26.4%
TurkeyLow sample: 4 decisions
50%
All judges 53.4%
Unknown NationalityLow sample: 3 decisions
0%
All judges 58.2%
Dominican RepublicLow sample: 3 decisions
33.3%
All judges 6.1%
LibyaLow sample: 3 decisions
100%
All judges 69.6%
NepalLow sample: 3 decisions
66.7%
All judges 62.3%
Sri LankaLow sample: 2 decisions
50%
All judges 44.9%
GuyanaLow sample: 2 decisions
50%
All judges 26%
KosovoLow sample: 2 decisions
100%
All judges 69.7%
MauritaniaLow sample: 2 decisions
100%
All judges 19.9%
AzerbaijanLow sample: 1 decision
100%
All judges 59.5%
BrazilLow sample: 1 decision
0%
All judges 16.2%
ChadLow sample: 1 decision
0%
All judges 36.2%
HaitiLow sample: 1 decision
0%
All judges 10.3%
HollandLow sample: 1 decision
0%
All judges 29.2%
IranLow sample: 1 decision
100%
All judges 58.7%
South KoreaLow sample: 1 decision
0%
All judges 14.9%
LiberiaLow sample: 1 decision
0%
All judges 24.8%
MaliLow sample: 1 decision
100%
All judges 34.3%
NigeriaLow sample: 1 decision
100%
All judges 47.9%
PeruLow sample: 1 decision
0%
All judges 12.3%
Saudi ArabiaLow sample: 1 decision
100%
All judges 65.6%
SpainLow sample: 1 decision
0%
All judges 6.9%
SudanLow sample: 1 decision
0%
All judges 43.9%
TajikistanLow sample: 1 decision
0%
All judges 46%
Soviet Union (former)Low sample: 1 decision
0%
All judges 84.5%
289 asylum decisions on the merits (grants and denials) since 2021, from 46 nationalities.

Grant rates by type of relief

Share granted among merits decisions for each application type, since 2021.
Asylum
55% 289 decisions
Withholding / CAT
11.8% 561 decisions
Cancellation — 42B (non-LPR)
31.8% Low sample: 22 decisions
Cancellation — 42A (LPR)
50% Low sample: 8 decisions
Adjustment of status (245)
91.7% Low sample: 12 decisions
Voluntary departure
72% 132 decisions

Bond decisions

Custody redetermination (bond) rulings since 2021. "Granted" means the judge lowered the bond, ordered release on recognizance, or otherwise granted the request.
65%
Bond requests granted
500 bond decisions
$4,000
Median bond when set
782 bonds with amounts
0.3%
Released on own recognizance
0.3% of grants
Also 178 requests denied as moot (nothing was left to decide), counted apart and left out of the grant rate.
This judge 65%
Lumpkin, GA, where these bonds were heard: 47.2% · median $6,000 (9,203 bond decisions)
National 50.3% · median $7,000

Bond in depth: amounts and outcomes by year

Every bond (custody redetermination) decision this judge entered since 2021, by calendar year of the decision (January to December; the last year runs to September 2026, when EOIR's data ends). Amounts are the bond the judge set when lowering it, raising it or granting amelioration, exactly as recorded. EOIR does not reliably record DHS's starting amount, so “lowered”, “raised” and “no change” come from the judge's decision code.
This judge 1,362 bond decisions · 782 with an amount set · median $4,000
Lumpkin median $6,000 (5,605 amounts set)

Amounts set, by range

Bonds this judge set, by amount range, with this court’s share
RangeThis judgeShareThis court
Under $1,50010.13%0.09%
$1,500 (statutory minimum)455.8%1.5%
$1,501–$5,00067887%46%
$5,001–$10,000536.8%38%
$10,001–$25,00040.51%13%
Over $25,00010.13%0.82%
Total782100%100%
The ranges follow the amounts judges actually set: $1,500 is the statutory minimum (INA § 236(a)(2)(A)), and $5,000 and $10,000 are the two most common amounts nationwide, so each sits whole at the top of its range.

Outcomes by year

This judge’s bond decisions by outcome and calendar year
Outcome2022202320242025All years
Granted
Bond lowered1046213316315
Released on own recognizance00101
Amelioration granted80019
Denied
No change to DHS’s bond8920037
No bond set4332593137
Amelioration denied10001
Not counted in the grant rate (including requests denied as moot)
Denied as moot10438342178
Bond raised37780110468
No jurisdiction (J)1227421192
No jurisdiction (D)00101
Request withdrawn27274623123
Bond decisions684275347561,362
Amounts set48414114215782
Median amount$4,000$4,000$3,000$3,000$4,000
Median at the court these bonds were heard at$5,000$5,000$5,000$6,000$6,000
Amounts set, by range
Under $1,50000101
$1,500 (statutory minimum)141017445
$1,501–$5,00043811511411678
$5,001–$10,000281510053
$10,001–$25,00031004
Over $25,00010001
* 2026 is a partial year: the data runs to September 2026.

Every amount set

Every bond amount this judge set, as recorded, by calendar year
Amount2022202320242025Total
$1,00000101
$1,500141017445
$2,00017820146
$2,50031105
$3,00013828403209
$3,50062008
$4,00018440313258
$4,50080008
$5,0008236224144
$6,0001184023
$7,00051107
$7,50020002
$8,00061209
$10,000453012
$12,00011002
$20,00020002
$80,00010001
Total48414114215782

In the courtroom

Languages in the courtroom

Spanish64%
English6.5%
Russian2.9%
Punjabi2.5%
Turkish2.5%
Language on the case record, 5,371 completed cases. An interpreter is provided by the court at no cost.

Custody

Detained when decided43%
Released from detention before the decision35%
Never detained22%
Custody status at completion, 5,398 cases.

Courts where this judge has decided cases

Hearing locations for cases this judge completed since 2021, by number of decisions. Many judges hear cases from more than one court (detained facilities, video dockets, or reassignments).
Lumpkin, GA · home court
3,360 (62%)
1,337 (25%)
421 (7.8%)
63 (1.2%)
58 (1.1%)
46 (0.85%)

If you have a hearing before this judge

The figures on this page describe past decisions, not yours. What you can change is how prepared you arrive: the right application filed on time, the evidence the judge will expect, and someone beside you who knows this courtroom.