Recommendation letters in an EB-1A petition are advisory opinions, not primary evidence. Under Matter of Caron International, USCIS may consider expert opinions but is not required to accept them — and under Matter of Soffici, opinions the record does not corroborate get less weight. Five to eight letters is typical, independent experts weigh most, and every claim should trace to an exhibit. This guide walks through a full annotated sample, the recommender mix that works, and what a letter can credibly say for each of the 10 criteria.
The legal status of recommendation letters is settled and widely misunderstood. Matter of Caron International, 19 I&N Dec. 791 (Comm'r 1988), holds that USCIS may, in its discretion, use advisory opinion statements from experts — but is not required to accept them, and may give less weight to an opinion that the record does not corroborate. Matter of Soffici, 22 I&N Dec. 158 (Assoc. Comm'r 1998), adds that going on record with unsupported assertions does not meet the burden of proof.
The practical consequence: a letter is only as strong as the exhibits behind it. A recommender who says "Dr. Zhao's paper has 510 citations" while Exhibit 14 shows the Web of Science record is supplying expert interpretation of corroborated fact — exactly what Caron International rewards. A recommender who says "Dr. Zhao is among the most influential researchers of her generation" with nothing behind it is supplying an unsupported assertion — exactly what Soffici discounts.
Five to eight letters is the practical norm for a competitive petition, with at least half from independent experts. Letters do not win EB-1A cases on their own; corroborated documentary evidence does. What letters do — when written by the right people about the right facts — is tell the officer what the evidence means: why this award is selective, why this citation count is exceptional, why this role is critical. That interpretive layer is what carries the Kazarian final merits determination.
Sources: Matter of Caron International, 19 I&N Dec. 791 (Comm'r 1988); Matter of Soffici, 22 I&N Dec. 158 (Assoc. Comm'r 1998); Kazarian v. USCIS, 596 F.3d 1115 (9th Cir. 2010); USCIS Policy Manual Vol. 6 Part F Ch. 2.
Independence is the axis USCIS weighs on. An expert who knows your work only by reputation sits at the top; a colleague down the hall sits at the bottom. Build the packet so at least half the letters come from the top of this list — and assign each letter different criteria, so the packet covers your strongest three or four criteria without redundancy.
Strongest
No employment, co-authorship, mentorship, or financial tie — the opinion cannot be attributed to loyalty or institutional interest. USCIS and AAO consistently give these letters the most weight. Best used for original contributions of major significance and the top-of-field opinion that supports the Kazarian final merits determination.
Strong
A CTO, chief scientist, or founder whose organization deployed, licensed, or benchmarked the petitioner's work testifies to real-world impact with numbers an officer can verify. Best used for leading or critical role, high remuneration context, and commercial significance of contributions.
Corroborative
Useful for first-hand facts no outsider can attest — who conceived the idea, what the petitioner's individual role was in a multi-author project. Not weighted as independent. Limit to one or two per packet, and never let collaborators argue acclaim.
Corroborative
Confirms employment record, project ownership, and role — facts that support the leading/critical role and high-salary criteria. The employment tie means the praise itself carries little independent weight; the verifiable facts are what count.
Weakest
A peer at the same employer shares the employer's interest in the petition outcome and rarely adds facts the employer letter does not. If included at all, the letter should attest to one narrow, specific fact — not opine on standing in the field.
A worked example from a fictional independent professor supporting two criteria — original contributions of major significance (criterion v) and judging the work of others (criterion iv) — plus the top-of-field opinion that supports Kazarian Step 2. Each paragraph carries a "why this works" annotation. Every name, institution, and number is invented; an adjudicator will recognize a copied letter immediately.
Robert Alvarez, PhD
Professor of Electrical Engineering & Holder, Whitfield Chair in Photonics
Georgia Institute of Technology
777 Atlantic Drive NW, Atlanta, GA 30332 · ralvarez@gatech.edu · (404) 555-0188
June 12, 2026
Re: EB-1A petition of Dr. Lin Zhao — independent expert opinion
To the Officer adjudicating this petition:
¶1 · Expert credentials
I am a Professor of Electrical Engineering at the Georgia Institute of Technology, where I hold the Whitfield Chair in Photonics. I have worked in integrated photonics for 22 years, published 180 peer-reviewed papers (h-index 64), and serve as an Associate Editor of Optica and as a Fellow of both IEEE and Optica (formerly OSA). I have chaired the technical program committee of CLEO, the field's flagship conference, and have reviewed grant proposals for NSF and DARPA photonics programs for over a decade.
Why this works: the recommender's authority is established with verifiable, named facts — chair title, h-index, editorships, fellowships, named conference. USCIS can confirm every claim from public sources, which is exactly the corroboration standard Soffici demands of the rest of the letter.
¶2 · Knows the petitioner by reputation only
I have never met, collaborated with, co-authored with, supervised, or been employed alongside Dr. Zhao. I know her work entirely by reputation: I first encountered her 2022 Nature Photonics paper on thermally stable microcomb sources when my own group attempted — unsuccessfully — to solve the same stabilization problem, and I have since cited her work in four of my publications. This letter is the unsolicited professional judgment of an independent expert.
Why this works: independence is stated explicitly and negatively (no met, no co-authored, no supervised), then anchored in a documentable channel — citations in the recommender's own papers, which appear in the exhibit record. An officer never has to take independence on faith.
¶3 · Specific, verifiable testimony — original contributions (criterion v)
Dr. Zhao's central contribution is the self-injection-locked microcomb architecture she introduced in 2022. Before her paper, no group — including mine — could hold a soliton microcomb stable across a 60°C ambient swing; her design holds lock across 85°C with no active cooling. The paper has accumulated 510 citations in under four years (top 0.5% in photonics per Web of Science, Exhibit 14), her reference implementation has been adopted by at least nine research groups on three continents (Exhibit 15), and two commercial LiDAR vendors have publicly credited the architecture in product documentation (Exhibit 16). This is an original contribution of major significance to the field: it converted microcombs from a laboratory curiosity into a deployable component.
Why this works: every number — 510 citations, 9 groups, 2 vendors, 85°C — is tied to a named exhibit. The letter interprets the evidence rather than substituting for it, which is precisely how Caron International says expert opinions earn weight. The closing sentence states the regulatory conclusion (major significance) in the recommender's own technical terms.
¶4 · Judging the work of others (criterion iv)
Dr. Zhao's standing is further reflected in how often the field asks her to judge the work of others. As documented in Exhibits 21-24, she has completed over 70 manuscript reviews for Nature Photonics, Optica, and Physical Review Letters since 2021; served on the CLEO technical program committee in 2024 and 2025, where committee members are selected by invitation from roughly the top one percent of active researchers; and was an invited judge for the Optica Foundation Challenge in 2025. Editors do not repeatedly assign reviews in this volume to researchers whose judgment the field does not trust.
Why this works: the recommender — himself a CLEO program chair and journal editor — is uniquely positioned to explain what the judging record means, not just that it exists. The exhibits prove the reviews happened; the letter supplies the significance an officer outside the field cannot infer alone.
¶5 · Comparison statement + top-of-field opinion (Kazarian Step 2)
In my 22 years in integrated photonics I have evaluated the work of well over a thousand researchers as an editor, program chair, and grant reviewer. Fewer than five of them have produced a single result that redirected the field's engineering assumptions the way Dr. Zhao's stabilization architecture did. In my professional judgment, she stands among the small percentage at the very top of the field of integrated photonics, and her acclaim — measured in citations, adoption, and invitations to judge — has been sustained, not episodic.
Why this works: the comparison statement ('fewer than five in over a thousand') gives the officer a concrete denominator for 'small percentage at the very top' — the exact Kazarian Step 2 language. Echoing the regulatory standard ('sustained acclaim') in a factual, quantified frame is what lets an officer quote this paragraph in an approval.
Closing · explicit support + contact for verification
For these reasons I support Dr. Zhao's EB-1A petition without reservation. I am available to USCIS to verify or expand on any statement in this letter; I can be reached directly at ralvarez@gatech.edu or (404) 555-0188.
Why this works: the offer to verify is not a courtesy — it signals the recommender stands behind every factual claim, which is the posture Caron International rewards. Institutional email and direct phone on letterhead make the letter checkable; notarization is not required and adds nothing.
Respectfully,
/signed/
Robert Alvarez, PhD
Professor and Whitfield Chair in Photonics, Georgia Institute of Technology
ralvarez@gatech.edu · (404) 555-0188
From a fictional industry recommender — a chief technology officer whose company recruited the petitioner. Note how the letter sticks to facts the recommender personally controls (org charts, compensation bands, business outcomes) instead of opining on academic standing.
Leading or critical role (criterion viii)
As Chief Technology Officer of Meridian Sensing (1,400 employees; Exhibit 31), I created the position of Principal Architect, Photonic Systems specifically to recruit Dr. Zhao in 2024. She leads the 38-engineer division responsible for our entire LiDAR product line, which generated $96M — 41% of company revenue — in fiscal 2025 (Exhibit 32). She reports directly to me, and the architecture decisions for both shipping products are hers. Her role is critical to this organization in the plain meaning of the word: without her division's output, our principal product line does not exist.
Why this works: 'leading or critical role' requires showing both the role and the organization's distinguished reputation — the letter quantifies the org (headcount, revenue share, reporting line) and ties each fact to an exhibit instead of asserting importance in the abstract.
High remuneration (criterion ix)
Dr. Zhao's total compensation in fiscal 2025 was $612,000 (Exhibit 33). As the executive who approves every compensation band in our engineering organization, I can attest this is the highest engineering compensation at Meridian and, per the Radford survey data we license (Exhibit 34), above the 99th percentile for photonics engineers in the United States. We pay it because her specific expertise is, in our direct hiring experience, not obtainable elsewhere at any price we tested.
Why this works: the recommender testifies only to what he personally controls — the comp approval — and points to third-party survey data for the comparison, because criterion ix turns on evidence of high salary relative to others in the field, not on the employer's opinion of it.
Each of these patterns appears repeatedly in AAO non-precedent EB-1A decisions where letter evidence was given little weight. Any one weakens a packet; several in combination are commonly fatal.
'Dr. X is a brilliant, world-class researcher' tells an officer nothing checkable. The most common RFE language on letters is that they 'do not specifically address the petitioner's individual contributions.' Every adjective should be replaced by a number, a name, or a comparison.
Adjudicators pattern-match across the packet. When three letters open with the same sentence structure or recycle the same phrases, all three are discounted at once — the AAO has done this to entire packets. Each recommender must write in their own voice.
A packet with no independent voices reads as the petitioner's own network vouching for itself. EB-1A turns on national or international acclaim — which by definition exists outside the petitioner's circle. At least half the letters should come from experts with no personal or professional tie.
Under Soffici, unsupported assertions do not meet the burden of proof, and under Caron International an officer may give uncorroborated opinions less weight. A citation count, deployment figure, or award mentioned in a letter but absent from the exhibits is a liability, not evidence.
A recommender who writes 'Dr. X meets criteria (iv), (v), and (viii) of 8 CFR 204.5(h)(3)' is doing the officer's job badly and their own not at all. Legal conclusions belong in the petition letter; the expert's value is field-specific facts and judgments only an insider can supply.
One line per criterion under 8 CFR 204.5(h)(3)(i)-(x). The pattern across all ten: the recommender testifies to facts and field context they personally know — never to legal conclusions, and never beyond what the exhibits support. See the full 10-criteria guide for the evidence each criterion requires.
Explain the award's selectivity from personal knowledge — how many candidates, who judges, which past winners went on to lead the field.
A current member or selection-committee veteran can describe the admission bar — election rates, who votes, what record is typically required.
Attest to the outlet's standing in the field and that coverage of an individual researcher's work at this level is rare.
An editor or program chair can explain how reviewers are chosen and what a high review volume signals about the field's trust in the petitioner.
The core letter territory: name the contribution, state what the field could not do before it, and quantify adoption — citations, implementations, products — against exhibits.
Contextualize venue prestige and citation impact — acceptance rates, field citation norms — rather than restating the publication list.
A curator or festival director can attest to the venue's selectivity and how the petitioner's work was chosen over the field of submissions.
A senior executive testifies to the org chart, the reporting line, and what business outcome depends on the petitioner — facts only the employer controls.
The compensation approver states the figure and points to third-party survey data for the field comparison — the criterion turns on relative position, not the raw number.
A promoter, label executive, or distributor attests to box office, sales, or streaming figures from their own books — numbers that must match the exhibits.
Five to eight is the practical norm, with at least half from independent experts. There is no statutory minimum. USCIS weighs independence and specificity over count — five strong, criterion-specific letters outperform a dozen generic ones from collaborators.
Recognized authorities whose own standing is publicly verifiable: tenured professors, senior industry leaders, journal editors, standards-body chairs, government scientists. The strongest letters come from independent experts who know your work by reputation only. Each letter should target different criteria.
Yes — but it counts as corroborative, not independent, because of the mentorship tie. Use it to confirm first-hand facts (who did the work, your specific role), include at most one or two such letters, and keep at least half the packet genuinely independent.
No. Notarization is not required and adds no evidentiary weight. What matters: signature, date, institutional letterhead, full title, and direct contact information so USCIS can verify the letter.
No employment, supervisory, mentorship, co-authorship, or financial relationship with you — past or present. The gold standard is an expert who encountered your work through normal professional channels (citing it, reviewing it, deploying it) without ever working alongside you.
Letters are advisory opinions, not primary evidence. Under Matter of Caron International, USCIS may use expert opinions but is not required to accept them; under Matter of Soffici, unsupported assertions do not meet the burden of proof. Every claim in a letter should map to an exhibit.
Yes — or at minimum substantially revise and personalize a structured draft in their own voice. USCIS detects boilerplate, and the AAO has discounted packets where letters share sentence structure. Brief the recommender with your CV and target criteria, then let them write.
No. Letters interpret the record; they cannot substitute for it. What wins cases is corroborated documentary evidence — award certificates, citation records, media coverage, judging invitations, salary data — with letters explaining why that evidence places you at the top of the field.
1.5-2 pages. One page reads thin; over three pages dilutes. Every paragraph should make a point USCIS can map to a specific criterion under 8 CFR 204.5(h)(3) or to the Kazarian final merits question.
Yes — auto-drafted starting points by recommender type, each mapped to the criteria that recommender is best positioned to address. Visacub does NOT ghostwrite letters recommenders sign as-is: the drafts are scaffolding the recommender must rewrite in their own voice. Visacub is self-help software, not a law firm.
The free assessment scores your record against all 10 criteria and shows which recommender types your case most needs. Visacub generates auto-drafted starting points by recommender type — recommenders revise in their own voice. Visacub is self-help software, not a law firm, and the sample on this page is fictional.