What does the 90th percentile wage mean in BLS OEWS data?

It is the wage below which 90 percent of workers in that occupation fall, so 10 percent earn more. It is not the top wage and not the maximum; it is a position in the distribution that experienced, high-performing workers in the occupation can reasonably target.

Is the median wage the same as the average wage in BLS data?

No. The median (50th percentile) is the middle value: half earn more, half earn less. The mean (average) can be pulled up by a few very high earners. In right-skewed occupations the mean sits above the median.

Which percentile should I use in a salary negotiation?

Early-career candidates typically land near the 10th to 25th percentile; mid-career near the median; experienced specialists can anchor near the 75th. The percentile that matches your experience level is your realistic target, not the median by default.

Where can I download BLS OEWS percentile wage data by metro area?

The BLS publishes OEWS data files with percentile wages by occupation and metro area each May. This site's free CSV compiles the metro-level figures into a single downloadable file for quick lookup.

Software Developer Salary by Metro Area: The May 2025 BLS Numbers

Q: What is the average software developer salary in the US?

BLS OEWS May 2025: mean $148,100, median $135,880 per year nationally for software developers (SOC 15-1252).

Q: Why do software salaries vary so much between cities?

Agglomeration and sorting. Research by Combes, Duranton, and Gobillon (2007) found worker sorting explains a large share of spatial wage gaps.

Q: Is BLS wage data accurate?

The most representative employer survey available, but wages only: no stock, bonuses, benefits, or self-employed workers.

Q: Should I use the mean or the median in a negotiation?

Median for positioning yourself, mean for the market's shape. Never mix one metro's mean with another's median.

BLS OEWS May 2025, software developers (SOC 15-1252) · updated October 2026 · 8 min read

Seattle pays software developers $174,920 a year on average. Las Vegas pays $129,940 at the median. Both numbers come from the same Bureau of Labor Statistics release, and comparing them directly is the single most common mistake in salary research. One is a mean, one is a median, and they answer different questions. This is the table I wish someone had handed me, with the numbers labeled honestly and the traps marked.

The source for everything below is the BLS Occupational Employment and Wage Statistics release for May 2025, published in May 2026. It is an employer survey covering about 830 occupations, and it publishes wages by metropolitan statistical area. When I say "metro," I mean the MSA: Houston means Houston-Pasadena-The Woodlands, not the city limits.

The numbers, labeled

Mean and median are not interchangeable, so the table keeps them separate. The mean answers "what does the market pay on average." The median answers "what does the middle developer earn." In tech the mean runs hotter because the top tail is long.

Seattle-Tacoma-Bellevue: $174,920 mean annual. The highest mean in this set, and it is not close.
Baltimore-Columbia-Towson: $153,910 mean annual. Defense and federal contracting money shows up here.
Washington-Arlington-Alexandria: $153,100 mean annual. Same story as Baltimore, larger market.
National, all metros: $148,100 mean annual; $135,880 median annual ($65.33/hour).
Las Vegas-Henderson-North Las Vegas: $129,940 median annual.
Houston-Pasadena-The Woodlands: $129,440 median annual.

The gap between Seattle's mean and Houston's median is about $45,000. Before you update your resume, read the next two sections. The number is real. What it means is subtler.

Why the ranking looks like this

Two forces, and economists have been arguing about their relative size for decades. The first is agglomeration: productive firms cluster together, and clustered firms pay more because they compete for the same people. The second is sorting: the most skilled developers move to the highest-paying metros, which makes the metro average look high even if no individual firm pays a premium for the location itself.

The sorting point is not hand-waving. Combes, Duranton, and Gobillon showed in "Spatial wage disparities: Sorting matters!" (Journal of Urban Economics, 2007, doi:10.1016/j.jue.2007.04.004) that once you account for worker skill differences across places, a large share of the spatial wage gap shrinks. Translation: Seattle does not just pay more. It employs a different mix of developers than Las Vegas does. If you are above the median skill in your current market, your personal number is higher than your metro's number, wherever you live.

The three traps in this table

Trap 1: BLS measures wages, not compensation. Stock grants, bonuses, and benefits are not in these numbers. At a public tech company the equity can exceed the base salary, which is why self-reported figures on sites like levels.fyi run much hotter than BLS. Use BLS for the market floor. Use total-comp data for the ceiling.

Trap 2: SOC 15-1252 is not your job title. BLS groups "software developers" into one occupation code. Your title might be frontend engineer, SRE, or "full-stack wizard." Close enough for market research, not exact enough to quote in a negotiation as if it were your role's number. Pair it with the percentile ladder: BLS OEWS Wage Percentiles Explained.

Trap 3: The metro is not the paycheck. A $45,000 gap between Seattle and Houston narrows fast once housing enters the picture. I wrote the full adjustment walkthrough separately: $95,000 in Columbus vs $140,000 in San Francisco: How to Actually Compare Job Offers Across Cities. Run the equivalent-salary math before you rank cities by the raw wage.

The question I cannot answer yet

Remote work should, in theory, flatten this ranking. If a Houston developer can earn a Seattle salary from a home office, the metro premium becomes a relic. In practice, most remote roles still carry location-based pay bands, and the BLS survey captures where the employer reports the job, not where the worker's laptop sits. The May 2025 data does not settle it. The next two releases might. I will update this table when they land, because this is the most interesting open question in salary data right now.

Compare the full dataset yourself.
Download the free BLS OEWS metro wage CSV and filter any occupation, any metro. No signup.
Download the free wage data

FAQs

What is the average software developer salary in the US?

In the BLS OEWS May 2025 release, software developers (SOC 15-1252) earned a mean of $148,100 and a median of $135,880 per year nationally. The median is the better "typical developer" number; the mean is pulled up by top earners.

Which metro pays software developers the most?

Among published BLS metro figures, Seattle-Tacoma-Bellevue leads this set at $174,920 mean annual. Coastal tech hubs consistently top the metro ranking, though the exact order shifts between releases and between mean and median measures.

Why do software salaries vary so much between cities?

Two forces: agglomeration (productive firms cluster and compete for talent, pushing wages up) and sorting (highly skilled developers concentrate in high-paying metros). Research by Combes, Duranton, and Gobillon (2007) found sorting explains a large share of spatial wage gaps.

Is BLS wage data accurate?

It is the most representative employer survey available, but it measures wages only: no stock, no bonuses, no benefits, and no self-employed workers. Treat it as the market floor for base pay, and check total-comp sources for the ceiling.

Should I use the mean or the median in a negotiation?

The median for positioning yourself ("I am at the 75th percentile, so..."), the mean for understanding the market's total shape. Never mix one metro's mean with another's median. The full walkthrough is in Median vs Mean Wage: Which BLS Number to Use in a Salary Negotiation.

One practical data guide a week. Get it by email.