Two new BIA precedent decisions landed this month. Both involve Mexican nationals. Both involve particular social groups. Both die on the same fundamental flaw — the group either defines itself by the harm, or it collapses for lack of any real definition at all. Read them together and a hard rule comes into focus.

Case One: The Domestic Violence Claim That Defined Itself Into Oblivion

A Mexican woman sought withholding of removal under section 241(b)(3) of the INA. She had endured sustained, documented abuse at the hands of a man with ties to the Los Gonzalez organized crime family — a man she described as controlling, violent, and connected enough to make flight within Mexico feel impossible. The Immigration Judge granted relief. The proposed particular social group: “married Mexican women who are unable to leave their relationship.”

The BIA reversed. The reason is structural and it cuts clean: the group defines itself by the persecution. “Unable to leave” is not an independent characteristic — it is the harm itself, restated as a membership criterion. The circularity is the defect. Under the framework the Board established in Matter of M-E-V-G-, 26 I&N Dec. 227 (BIA 2014), and reinforced through Matter of A-B- I and II, a cognizable particular social group must be defined by characteristics that exist independently of — and prior to — the harm the applicant claims to fear. A group whose membership requirement is the persecution itself cannot satisfy that standard. The IJ had relied on Matter of A-R-C-G-, the 2014 precedent that had briefly opened the door to DV-based PSG claims. The current Attorney General vacated that decision and replaced it through Matter of S-S-F-M-, 29 I&N Dec. 207 (A.G. 2025), reinstating the A-B- framework in full. Leaning on buried precedent was fatal.

The rule: a particular social group must exist before the persecution — not because of it. You cannot build the group around vulnerability to the very harm you fear.

The BIA reversed on three additional grounds, each of which should matter independently to practitioners building DV-based claims.

On marriage: cohabitation and having children in common do not establish a lawful marriage. The validity of the marital relationship must be determined by the law of the place and time of the alleged union — not by the fact of shared residence or shared parenthood. The IJ presumed a common-law marriage from those facts alone. That analysis was legally insufficient. If your group definition turns on marital status, you must build a record that addresses Mexican law on marriage formation, and you must establish that the specific relationship at issue meets that standard.

On government protection: the applicant claimed her former husband had law enforcement connections that would neutralize any attempt to seek help. The Board found that assertion speculative and unsupported by credible evidence. The State Department country report cut the other direction — Mexico investigates, prosecutes, and incarcerates domestic abusers. An IJ cannot find government unwillingness to protect based on unsubstantiated claims about an abuser’s local connections while ignoring a country report documenting active prosecution of domestic violence.

On internal relocation: the applicant never attempted to relocate within Mexico. She owned a car. She had work experience including self-employment. Nothing in the record established that her former partner had the reach or the motivation to track her across the country. The IJ’s conclusion that relocation was unreasonable was not supported by the evidence.

The CAT claim was never adjudicated — the IJ denied it as moot after granting withholding. The case remands on CAT alone. That door stays open, but it is a narrow one given the record.

Case Two: A Diagnosis Without Defined Symptoms Is Not a Group

A Mexican national — found incompetent to represent himself, provided a qualified representative, and carrying one of the more extensive diagnostic profiles to appear in a BIA opinion — sought asylum on the ground that he would be institutionalized and persecuted in Mexican psychiatric facilities. His diagnoses included schizoaffective disorder, bipolar disorder, major depressive disorder, PTSD, intellectual disability, major neurocognitive disorder, delusional disorder, and unspecified anxiety disorder. The IJ found his proposed group cognizable and his fear of future persecution well-founded. The proposed group: “Mexican men with Schizoaffective Disorder.”

The BIA reversed under the three-part M-E-V-G- framework — immutable characteristic, particularity, social distinction. Mental illness can satisfy immutability; the Board has acknowledged that. But the proposed group collapsed at particularity. Schizoaffective disorder is not a fixed, uniform condition. It covers a wide spectrum of severity and symptom expression. It varies over time within the same individual. Medication can suppress symptoms entirely or substantially. Under the Board’s own precedent and consistent circuit court authority — including Andrade v. Garland, 94 F.4th 904 (9th Cir. 2024), and Temu v. Holder, 740 F.3d 887 (4th Cir. 2014) — a diagnosis that sweeps across that range without limiting language does not have the definable boundaries the INA requires.

The IJ had identified symptoms — grandiose and persecutory delusions, auditory and visual hallucinations, disorganized thinking and behavior — but those symptoms appeared nowhere in the proposed group definition, were not established in the record, and had not even been observed while the respondent lived in Mexico, where he went undiagnosed and unmedicated for years. A respondent must define the specific symptoms, establish that the group is bounded by them, and then prove that he actually exhibits them. None of that happened here.

The rule: a diagnosis is a starting point, not a destination. Define the symptoms, establish that the respondent exhibits them, build that into the group — or the group dies at particularity.

On social distinction, the BIA drew a line that matters: the question is not whether a clinician distinguishes schizoaffective disorder from bipolar disorder. The question is whether Mexican society perceives individuals diagnosed with schizoaffective disorder — as a group defined by that diagnosis — as socially distinct. The IJ relied on generalized country conditions evidence documenting stigma against mentally ill people broadly. That is not enough. Mexican society may respond to erratic or bizarre behavior, but that behavioral profile crosses diagnostic lines — intoxication, drug use, bipolar disorder, and various other conditions can produce the same visible symptoms. Society does not parse the DSM-5 on the street. The social distinction analysis must track the specific group, not mental illness as a genus.

Even setting cognizability aside, the Board rejected the fear of persecution as a five-link speculative chain: inability to access medication → erratic behavior → contact with authorities → institutionalization → persecution inside. No link carried independent evidentiary support. The respondent lived in Mexico without medication and without incident from 2006 to 2022. His encounters with U.S. law enforcement stemmed from drug use, not psychiatric symptoms. His family had helped him access and afford medication for his HIV diagnosis in Mexico — a fact that undermined his claim that medication would be unavailable. And critically, the Board applied well-settled law: poor conditions in a psychiatric institution — understaffed, undertrained, under-resourced — do not constitute persecution without evidence of persecutory motive. Inadequate healthcare is not a nexus to a protected ground. Neglect is not persecution.

CAT remands here as well. The IJ never reached it, having granted asylum. That question remains open on remand.

The Common Thread

Both decisions enforce the same structural demand on every particular social group claim that reaches the BIA. The group must be defined with precision — bounded by characteristics that exist independently of the harm, supported by a record that answers hard questions about membership, and recognized by society in a way that doctors or lawyers alone cannot supply.

A diagnosis without defined symptoms is not a group. Vulnerability to harm is not a group. Neither satisfies the standard. The BIA has now said so twice in the same month, in binding decisions that reach every Immigration Judge in the country.