H-1B Salary Database Guide 2026: How to Search, Read & Compare Real LCA Salaries

Every H-1B salary database on the internet is built from the same public source: the U.S. Department of Labor's Labor Condition Application (LCA) disclosure data. The databases differ in how they present it — but the numbers that matter are the same ones. This guide teaches you to read those numbers properly: what an LCA wage actually is, what wage levels I–IV mean, how to search by company, job title and state, and how to compare two offers without fooling yourself. Every figure below comes from real DOL records through VisaTalentUSA's own database — no estimates, no dummy data.

What an H-1B salary database is

An H-1B salary database is a searchable index of Labor Condition Applications. Before an employer can file an H-1B petition with USCIS, it must first file an LCA with the Department of Labor and get it certified. Each certified LCA becomes a public record, and salary databases collect those records so you can search them by employer, job title, or worksite state.

One LCA record typically contains the employer's name, the job title, the occupation code, the worksite city and state, the wage offered for the position (usually as an annual figure), the prevailing wage for that occupation and area, the prevailing-wage level (I–IV), whether the job is full-time or part-time, the employment dates, and the case status — Certified, Certified - Withdrawn, Denied, or Withdrawn. The wage you see in a salary database is the offered wage from that record.

Two things follow from this. First, a salary database shows offered wages, not payroll data — nobody's payslip is in it. Second, filing counts are counts of applications, not counts of people: a single LCA can cover more than one worker, so “19,937 filings” does not mean 19,937 individual people.

Where the data comes from

The source is the Department of Labor's Employment and Training Administration, which publishes LCA disclosure files for every fiscal year. VisaTalentUSA normalizes those files through an open-source pipeline into a searchable database: 1,532,089 H-1B records across FY2024–FY2026. Employer names and job titles are grouped and normalized — for example, “Sr. Software Engineer” rolls into “Software Engineer” — so counts reflect real demand, not spelling variants.

The sponsor side of the site covers the Top 1,000 employers — the largest H-1B filers, which is the most useful slice for job targeting, not an exhaustive list of every filer. New employer records are added daily, so published totals move slightly over time; the worked examples in this guide are FY2026 figures through Q3, checked on October 28, 2026. The title side covers hundreds of job titles, each with its own filing count, median offered salary, and top employers.

If you only remember one thing from this section, make it this: every salary figure on this site traces back to a DOL record. When a figure looks surprising, the explanation is almost always in how the record was filed — the wage level, the worksite, or the occupation — not in an error in the database.

LCA wage vs actual salary: the difference that confuses everyone

The single most misunderstood number in any H-1B salary database is the wage itself. On the LCA, the employer attests that it will pay at least the prevailing wage for the occupation and area, and at least the actual wage it pays similarly employed workers — whichever is higher becomes the required wage. The wage written on the LCA is the offered wage for that position at the time of filing.

Your actual salary can still differ from the LCA wage, for honest reasons:

• Bonuses and equity are not in it. The LCA wage is base pay for the position. Stock, bonuses, and signing bonuses do not appear in the disclosure, so total compensation is usually higher than the LCA figure at large tech employers.
• It is frozen at filing time. Raises, promotions, and role changes after the LCA is filed do not update the old record.
• It is the job's rate, not your personal deal. Two people in the same title at the same company can have different LCAs at different wage levels.
• Part-time work annualizes oddly. A part-time hourly rate converted to an annual figure is not directly comparable to a full-time salary.

The practical rule: treat the database median as the middle of the market for that employer or title — a negotiation benchmark. If your offer sits well below the median for the same title and wage level, that is a real signal worth raising. If it sits above the median, the LCA figure was never a ceiling.

Prevailing wage levels I–IV: what each level means

Every LCA carries a prevailing-wage level, set by the job's requirements — education, experience, and supervision. The level is not a performance rating; it describes how demanding the position is, and it drives the prevailing wage the employer must meet.

Levels matter twice over. First, a higher level means a higher prevailing wage, so comparing a Level I salary with a Level IV salary for the same title is meaningless — always compare within the same level when you can. Second, under the wage-weighted H-1B lottery selection rule, higher wage levels receive more entries in the selection, so your level affects both pay and lottery odds. You can estimate that effect with the wage-level tool and the lottery odds calculator.

A real example makes the mix concrete. Amazon's FY2026 (through Q3) filings show the following wage-level mix, among the 15,250 cases that carry a wage level on file:

Note the denominator: 4,687 of Amazon's 19,937 FY2026 (through Q3) cases carry no wage level in the record, so the shares above describe only the 15,250 cases with a level on file. That is normal in LCA data — a missing field is reported as missing, never filled in with a guess.

Step by step: search by company

Company search answers one question: does this employer actually sponsor, at what volume, and at what pay? Work it in four steps, using Amazon as the example:

1. Open the sponsor profile. From the Top 1,000 sponsor list, open Amazon's profile.
2. Read the three headline numbers. For FY2026 (through Q3), Amazon filed 19,937 LCAs, of which 19,933 were certified — an LCA certification rate of 99.98% — with a median offered salary of $158,000 and an average of $159,696. Volume tells you the program is real; the certification rate tells you the filings are clean; the median tells you the market level.
3. Read the top titles. Amazon's largest FY2026 (through Q3) title groups are Software Dev Engineer (6,137 filings, $163,834 average), Software Development Engineer (2,459 filings, $168,352 average), and Manager (1,091 filings, $202,650 average). If your title is not in the top list, the company may still sponsor for it — check the title search instead.
4. Read the worksite states. The profile breaks filings down by where the work is performed (see the state section below). A company can sponsor heavily in one state and barely at all in another.

Repeat the same four steps for any second employer you are considering, and you have a like-for-like comparison — same year, same metric, same source.

Step by step: search by job title

Title search answers the reverse question: which employers hire people with your exact job title, and what do they pay? Use the Sponsor Match Finder:

1. Type your title. Start typing — for example, “Software Engineer”. The search suggests only titles that actually appear in DOL records, so a typo gives you a “try a broader term” hint instead of a dead end.
2. Read the snapshot. For Software Engineer in FY2026 (through Q3): 40,003 filings, a median offered salary of $151,000, and 39,360 certified cases — an LCA certification rate of 98.4%. The filing count is your demand signal; the median is your negotiation benchmark.
3. Read the top employers. For Software Engineer, the largest FY2026 (through Q3) filers are Google (4,779), Meta (Facebook) (2,439), Walmart (892), LinkedIn (795), and PayPal (651). Those are the companies with proven, active programs for your exact title.
4. Open the full title report. Every title has a dedicated report page with the full employer table and salary context. Bookmark it and use it to shortlist where to apply.

Title search is also how you catch a common mistake: searching a fancy internal job title (“Software Development Engineer II”) instead of the normalized title the database uses. If your exact phrase finds nothing, try the broader base title — the grouping exists precisely so spelling variants do not hide real demand.

Step by step: search by state

State search has one honest complication: VisaTalentUSA has no standalone state-search page, because a state number on its own misleads — the same employer pays very different wages in different states, and a state total mixes every occupation together. The correct way to answer a state question is through sponsor profiles, which break each employer's filings down by worksite state.

Amazon again, FY2026 (through Q3): 10,310 of its 19,937 filings (51.7%) list Washington as the worksite state, followed by California (2,995), Texas (2,123), New York (1,242), and Virginia (1,154). So “what does Amazon pay in Texas?” is a meaningful question — but “what does Texas pay?” is not, until you fix the employer and the title too.

Two cautions for state numbers. First, worksite state is where the work is performed, not where headquarters sits — never read a state split as a headquarters list. Second, a state with few filings gives a noisy median: a 50-filing state median moves on a handful of cases, while a 10,000-filing state median is stable. Weight your confidence by the filing count, which is always shown next to the figure.

How to compare two offers using the data

The only rule that matters: compare the same metric, in the same year, from the same source. A title median is not comparable to an employer median, and a full-year figure is not comparable to a through-Q3 figure. With that rule, two tables do most of the work.

First, employers — FY2026 (through Q3) median offered salary, all titles:

Second, job titles — FY2026 (through Q3) median offered salary, all employers:

Use the tables like this. If you hold an offer from Cognizant at $115,000, the employer table tells you that sits just above Cognizant's own $111,030 median — a fair offer by that company's own market, even though it trails Google's $193,000. If you hold a Software Engineer offer at $140,000, the title table tells you it trails the $151,000 title median — a real, citable data point for negotiation. Neither table promises you a number; both tell you where your number stands.

Honest limits of the data

A salary database is only useful if you know where it stops. These limits apply to every figure in this guide:

• FY2026 is partial. All FY2026 figures here are through Q3 (data ends June 30, 2026). Never compare a through-Q3 figure with a full-year figure, and only compare year-on-year for the same period.
• Certified - Withdrawn counts as certified. The LCA certification rate counts both “Certified” and “Certified - Withdrawn” as certified. Counting only “Certified” would understate real certification rates sharply.
• Medians are offered wages, not total compensation. Equity, bonuses, and benefits are not in LCA data. A lower LCA median at a company with heavy equity pay can still mean higher total pay.
• Top 1,000 is a slice, not the whole market. The sponsor ranking covers the largest filers. Smaller sponsors with a handful of filings exist in the full DOL files but are not the focus of the ranking.
• Figures move as records are added. New employer records are added daily, so a median checked today can shift slightly tomorrow. That is the data improving, not an error.
• The certification rate is a compliance signal, not a hiring promise. A high LCA certification rate means clean DOL filings. It does not mean the company will hire you, and it is not the USCIS petition approval rate.

Frequently asked questions

Is the salary data in H-1B databases real?

Yes — when the database cites the Department of Labor. Every figure in this guide comes from DOL Labor Condition Application disclosure data: 1,532,089 H-1B records across FY2024–FY2026, normalized through an open-source pipeline. If a salary site cannot name its source, treat its numbers as marketing, not data.

What is the difference between LCA wage and actual salary?

The LCA wage is the offered wage the employer attests on the Labor Condition Application at filing time — it must be at least the prevailing wage and at least the actual wage paid to similarly employed workers. Actual salary can be higher because bonuses, equity, raises after filing, and role changes are not in the LCA record. Use the database median as a market benchmark, not as a prediction of your payslip.

What do wage levels I–IV mean?

They are prevailing-wage levels set by the job's requirements: Level I is entry-level work, Level II is qualified work, Level III is experienced work, and Level IV is fully competent, expert-level work. Higher levels carry higher prevailing wages, and under the wage-weighted lottery rule they also receive more selection entries. Always compare salaries within the same level when you can.

How do I find salaries for my job title?

Use the Sponsor Match Finder: type your title and read the snapshot — total FY2026 (through Q3) filings, median offered salary, and LCA certification rate — plus the ranked list of top sponsoring employers for that exact title. For Software Engineer, that snapshot is 40,003 filings and a $151,000 median.

How do I find salaries at a specific company?

Open the company's sponsor profile from the Top 1,000 sponsor list. For example, Amazon's profile shows 19,937 FY2026 (through Q3) filings, a $158,000 median offered salary, a 99.98% LCA certification rate, and breakdowns by top titles and worksite states.

Can I compare two job offers with this data?

Yes, if you compare the same metric in the same year: employer medians against employer medians, or title medians against title medians, all FY2026 (through Q3). An offer above the matching median is strong; an offer well below it is a citable negotiation point. Never compare a title median with an employer median, and never compare a through-Q3 figure with a full-year figure.

Is the median salary a guaranteed salary?

No. The median is the middle offered wage across filings for that employer or title — half the filings are above it, half below. Your own offer depends on your wage level, worksite, experience, and negotiation. Treat the median as the market level, not as a promise.

What does the DOL LCA certification rate mean?

It is certified LCA cases divided by filed cases, where “certified” includes both “Certified” and “Certified - Withdrawn”. It measures DOL paperwork certification — a compliance-quality signal. It is not the USCIS H-1B petition approval rate, and a high rate does not guarantee any individual petition will be approved.

Sources

• U.S. Department of Labor, Employment and Training Administration — Foreign Labor Certification performance data and disclosure files
• U.S. Department of Labor — Prevailing wages and wage levels (OFLC wage guidance)
• USCIS — H-1B Specialty Occupations (petition process overview)

Published October 28, 2026. All salary figures are FY2026 (through Q3) medians and counts from U.S. Department of Labor LCA disclosure data via VisaTalentUSA's database, checked October 28, 2026. Figures move slightly as new employer records are added daily.