Every time you look up "what does a job make," you're relying on data that came from somewhere - and where it came from determines whether you can trust it. A number scraped from self-reported survey entries on a jobs site is a very different thing from an official government wage estimate built on responses from hundreds of thousands of employers. Understanding the difference, and knowing how the most authoritative source actually works, lets you read any salary figure with the right amount of confidence. This guide explains where reliable U.S. salary data comes from, how it's collected, and the limits worth keeping in mind.

The gold standard: BLS and the OEWS program

In the United States, the most authoritative wage data comes from the Bureau of Labor Statistics (BLS), a federal agency, through a program called the Occupational Employment and Wage Statistics (OEWS) survey. It is the source that most credible salary tools, news articles, and government reports ultimately draw on, including the salary figures on this site.

What makes OEWS the benchmark is its scale and method. Rather than asking individuals to type in their own pay - which invites errors, exaggeration, and selection bias - OEWS surveys employers directly about what they pay their workers. Over a multi-year cycle it collects responses covering millions of workers across hundreds of industries, then produces wage estimates for around 800 occupations in every state and hundreds of metropolitan areas. Because it's built from payroll reality reported by the businesses themselves, it avoids the biggest distortions that plague self-reported salary sites.

How the data is collected

OEWS works on a rolling survey. The BLS gathers wage reports from a large sample of establishments, weights the responses so they represent the full workforce rather than just who happened to reply, and combines several periods of data to build stable estimates. The results are then broken down two ways that matter for anyone comparing pay:

  • By occupation, using a standardized classification system so that "registered nurse" or "software developer" means the same thing everywhere the data is compared.
  • By geography, down to the metropolitan-area level, which is what makes it possible to see how the same job pays in one city versus another.

For each occupation in each area, OEWS reports not just a single number but a distribution - the median, the mean, and wages at the 10th, 25th, 75th, and 90th percentiles. That range is what lets you see both typical pay and how much it varies. (If percentiles are new to you, our guide on salary percentiles breaks down what each one means.)

Why this beats self-reported salary data

Plenty of popular salary tools rely on numbers users type in themselves. That approach has real weaknesses: people who volunteer their salary aren't a random sample, some round up, some report total compensation as base pay, and small job categories can rest on a handful of entries. The result can be figures that drift from what a typical worker actually earns.

Employer-reported government data sidesteps most of that. The sample is large and structured, the definitions are consistent, and there's no incentive to inflate. It won't match any single person's paycheck - no average does - but as a picture of what a job pays across a whole market, it's the most trustworthy starting point available. That's precisely why it's the foundation here rather than crowd-sourced entries.

The limits worth knowing

No data source is perfect, and using BLS numbers well means understanding what they can't tell you.

They lag the present. Official estimates are built from surveys collected over time and released on a schedule, so they reflect a reference period that may be a year or more before you read them. In a fast-moving field or a period of rapid wage growth, current market pay can run ahead of the published figure. Treat the number as a well-grounded baseline, not a live quote.

They're broad averages, not your offer. OEWS describes an occupation across a whole area. Your specific pay depends on your employer, industry, years of experience, specialization, and negotiation. Two people with the same job title in the same city can earn very different amounts, and both can be "normal."

Occupation categories are wide. A single job code can cover a range of real-world roles with different pay. "Software developers," for instance, spans many specialties and seniority levels bundled into one estimate.

Some cells are suppressed. Where a sample is too small to produce a reliable figure, BLS withholds it for data-quality reasons - which is why some occupation-and-city combinations simply have no published wage. On this site, we don't invent a number in those cases; we just don't show a page for them.

How to use BLS data wisely

Read government wage data as a reliable anchor, then adjust for your own situation. Use the median to gauge typical pay, use the 10th-to-90th range to see where you might fall, and remember that the figure is a baseline that current offers can exceed. Above all, pair the salary with local cost of living - a wage means something different in every city, which is the whole point of looking at pay and prices together. You can see this in action by comparing an occupation like registered nurses or software developers across metros, and our guide on nominal vs. real wages explains why the same BLS median translates into very different living standards depending on where you earn it.

The takeaway

The salary figures worth trusting come from employer-reported government data - the BLS OEWS program - not from self-reported entries on a jobs board. That data is authoritative because of how it's collected: at scale, from businesses, with consistent definitions and geographic detail. But it lags the present, describes broad averages rather than your specific pay, and occasionally leaves gaps. Read it as a solid baseline, adjust for your circumstances and your city, and you'll get an honest, well-grounded picture of what a job really pays.