H-1B lottery gaming, from the filings: the rate tripled to 25%
In 2024, Bloomberg revealed that thousands of companies — mostly small IT-staffing firms — were gaming the H-1B lottery by registering the same worker through several related entities, so that person got many entries while everyone else got one. It estimated roughly one selected registration in six was won this way. USCIS agreed: multi-registrations had jumped from 165,180 to 408,891 (+147%) in a single year, and it rewrote the lottery to select by person rather than by registration. That's the known story, and it's worth knowing.
Working from the same public source — the FOIA-released I-129 petition microdata — we rebuilt it from the ground up, and three things the pool-level headlines don't show fall out of it: how fast the gaming rose among the petitions that were actually filed (a conservative lower bound), a clean employer size-gradient that puts numbers on Bloomberg's "small firms" observation, and the India×IT-services interaction underneath it. Everything below is reproducible from the public data.
New: the gaming rate among the petitions that were filed
The counts everyone cites are registration-pool numbers. Restricting to the petitions that were selected and filed, the share tied to a multi-registered beneficiary still tripled in four years — 8.2% (FY2021) to 25.5% (FY2024):
| Fiscal year | Filed petitions | Multi-registered | Rate |
|---|---|---|---|
| 2021 | 99,610 | 8,148 | 8.2% |
| 2022 | 89,535 | 15,159 | 16.9% |
| 2023 | 91,832 | 20,892 | 22.8% |
| 2024 | 91,864 | 23,401 | 25.5% |
USCIS flags a beneficiary as multi-registered (ben_multi_reg_ind = 1)
when the person appeared in more than one employer's registration that lottery year.
It is USCIS's marker for potential abuse — not an adjudication of fraud,
and the employer that ultimately filed is not necessarily the party that
multi-registered the person.
A behavioral change, not a composition shift
India's share of filings was essentially flat (64% → 67% → 70% → 67%), so the rise isn't a country-mix effect. The rate rose within every origin group:
| Multi-reg rate among filed petitions | FY21 | FY22 | FY23 | FY24 |
|---|---|---|---|---|
| India-born | 10.5% | 23.4% | 30.0% | 34.9% |
| China-born | 5.4% | 4.6% | 8.1% | 10.1% |
| All other | 2.9% | 2.9% | 3.9% | 4.2% |
The real cut: India × IT-services staffing
A bare "India 24.6% vs China 6.9%" is true but under-specified. The phenomenon is an India × IT-staffing interaction (NAICS 5415, computer-systems-design services), pooled FY21–24:
| IT services (NAICS 5415) | Other industries | |
|---|---|---|
| India-born | 33.5% (n=156,643) | 9.7% (n=93,082) |
| Other-born | 6.9% (n=20,543) | 4.6% (n=102,573) |
India-born beneficiaries outside IT services run ~10% (still double other countries); India × IT-services reaches 45.4% by FY2024. The sector is half the story.
Quantifying Bloomberg's "small firms": a clean size-gradient
Bloomberg's reporting pointed at small staffing shops rather than the household-name outsourcers; the microdata turns that observation into a clean gradient. Multi-registration rates among the biggest filers are low — Amazon 3.6%, Infosys 3.1%, TCS 4.7%, Cognizant 4.5%, Microsoft 3.3%, IBM 3.1%.
The extreme rates are tiny staffing firms: Aclat Inc. 88 of 88 petitions (100%), Snowstack LLC 96.8%, R2 Technologies 91.9%. A clean size gradient, employers bucketed by total FY21–24 filings:
| Employer size (filings) | Employers | Petitions | Multi-reg rate |
|---|---|---|---|
| 1–9 | 54,807 | 105,504 | 20.4% |
| 10–49 | 4,656 | 91,430 | 35.2% |
| 50–199 | 551 | 46,972 | 19.8% |
| 200–999 | 98 | 38,537 | 4.4% |
| 1,000+ | 29 | 90,398 | 3.2% |
1,815 employers with at least 10 filings and a 50%-or-higher multi-reg rate account for 46% of all multi-registered petitions. Employers filing fewer than 50 petitions hold 79% of them. The gaming lived in thousands of small staffing firms, not the household-name outsourcers.
The pool was worse than the filings show
Our number counts filed petitions, which understates the lottery pool: many multi-registered beneficiaries were selected but never filed. USCIS's registration-level statistics:
| Cap FY | Eligible registrations | Multi-reg share of registrations |
|---|---|---|
| 2021 | 269,424 | 10.4% |
| 2022 | 301,447 | 29.9% |
| 2023 | 474,421 | 34.8% |
| 2024 | 758,994 | 53.9% |
The pool share rose faster (10.4% → 53.9%, 5.2×) than our filed-petition rate (8.2% → 25.5%, 3.1×) — so selection bias makes our figure conservative, not inflated.
What changed next: the FY2025 beneficiary-centric rule
This is the "before" picture. Starting with the FY2025 lottery, USCIS switched to beneficiary-centric selection — each person is entered once no matter how many employers register them, so running one beneficiary through several shells no longer improves their odds. In the first year under the rule, multi-registered registrations fell 408,891 → 47,314 (−88%) and total eligible registrations dropped from 759k to 470k.
How much of that is the rule? The multi-registration collapse tracks the rule's mechanism closely: once duplicate entries stop paying off, the reason to file them is gone. But the overall pool didn't shrink in a vacuum, and the raw before/after is not a clean natural experiment. Several things moved at once in 2024–25: the per-registration fee later jumped from $10 to $215 (FY2026), the tech-hiring market cooled, and enforcement scrutiny rose — and the pool kept falling the next year (470k → 344k) well after the counting change. So the defensible read is narrow: the rule removed the specific incentive this story documents and multi-registration cratered right after; the broader decline in how many people register reflects several 2025 changes together, not this one measure.
How to read these numbers
- Every rate is "among selected-and-filed petitions" unless labeled as the registration pool.
- Aggregate only — individual "one person → N shell companies" chains cannot be reconstructed (FOIA-redacted keys).
- "Multi-registered" is USCIS's flag for potential abuse, not proven fraud. The named high-rate employers are shown with their petition counts for that reason.
- Coverage: FY2021–2024, cap-subject lottery petitions, frozen snapshot.
Explore the employers yourself
Look up any H-1B filer in the employer directory or the sponsor rankings — each employer profile shows that company's I-129 approval rate and how its reported pay compares to the posted wage. You can also search the certified-wage records by employer, role and state.
Source & method
The petition-level figures come from the I-129 microdata released under the Freedom of Information Act — "sourced from USCIS, obtained by Bloomberg" (Bloomberg v. DHS), the same dataset behind Borjas's H-1B wage-gap paper (NBER w34793). Registration-pool figures (eligible registrations, selections, the FY2025 rule effects) are from USCIS's own published H-1B Electronic Registration statistics and its FY2025 selection-rule rulemaking.
Universe: fiscal years 2021–2024, cap-subject H-1B lottery petitions that were selected and filed (372,841 rows) — not the full registration pool. A frozen FOIA snapshot; per-beneficiary identifiers are redacted, so figures are aggregates only. Every number on this page was checked against the live database on 2026-07-13.
Custom data slices for research or reporting. If you're a researcher, academic or journalist and want a specific cut of this data — a breakdown by employer, occupation, country of birth, fiscal year, or any combination not shown here — get in touch and I'll put it together. Attribution to visa-bulletin.us is appreciated.