The Bureau of Labor Statistics expects 39,300 paralegal openings a year between now and 2034. Ten years of that is roughly 393,000 seats to fill. Net new jobs over the same decade: 600.
Not 600 a year. Six hundred, total. The occupation goes from 376,200 people in 2024 to 376,800 in 2034, which rounds to zero percent growth against 3 percent for the workforce generally, and along the way it will hire something close to its entire current headcount over again. That is the treadmill. Every year the profession replaces about a tenth of itself, and after ten years of running it is standing in exactly the same spot.
The number is a floor, not an estimate
BLS counts an opening only when somebody leaves the occupation. Its own definitions page says a worker who changes jobs within an occupation does not create one, because nothing net has happened. Nobody entered the profession and nobody left it. One chair got swapped for another.
So the paralegal who quits you on a Friday and starts at the firm across town on Monday is not in the 39,300 anywhere, because as far as the projection is concerned nothing happened.
That's the whole distance between an interesting statistic and a number you can budget against. Your actual replacement rate for that seat is 10.4 percent a year plus every firm to firm move in your city, and BLS has no visibility into the second part and never will. Ten point four is the floor. The real figure is worse and nobody knows by how much.
They are not retiring
I went in expecting paralegals to churn faster than everybody else. They don't. Separations run 10.4 percent a year against 10.6 percent for all occupations, so slightly under average. If you wanted a statistic proving the job burns people out at an unusual rate, it isn't in the data, and I would rather say that than not.
The mix is a different story. BLS splits leavers into people who go do a different kind of work and people who drop out of the labor force. Paralegals transfer out at 7.2 percent against a 5.9 percent national average, and they exit the labor force at 3.2 percent against 4.7. Seven in ten paralegal departures are somebody deciding to do something else for a living.
The seat isn't emptying because people are aging out of it. They're leaving the career, and you can't pay your way out of somebody concluding this isn't the one.
Now the seats nobody measures at all
Paralegal is the only role in your building you can look up. That's the strange part of this.
At budget time the actual question is whether what you pay your intake specialist is normal, and there's no answer to it. Robert Half's 2026 guide prices 39 legal titles and not one of them is intake or case management. NALP covers attorneys. The ALA survey runs past 400 pages and gets all the way through docket clerks and conflicts analysts without ever landing on the two jobs a contingency practice hires most often.
They aren't bad surveys. They're built for a different firm than yours. And the tell isn't that they lean defense side, which is what I assumed going in, it's that they don't know the distinction exists at all: the word plaintiff does not appear in Robert Half's guide.
The classification system has the same hole one level down. Lawyers are 23-1011. Paralegals are 23-2011. Legal secretaries sit over in 43-6012. There is no code anywhere in the federal system for a legal case manager, and the closest thing carrying the name is 21-1022, healthcare social workers, which is a different job in a different major group done by different people. So the role falls into 23-2099, Legal Support Workers, All Other, the residual bucket, where pay runs from $41,510 at the tenth percentile to $176,950 at the ninetieth. That isn't a finding about a job, it's a drawer with several unrelated jobs in it. The job boards will still quote you a figure for the title, and the three I checked disagreed by nearly two to one, so don't use them.
What to pay for the one you can look up
Use the industry number, not the headline. The paralegal median quoted everywhere is $61,010 as of May 2024, but that's across all industries and it's pulled upward by federal government paralegals at $77,940. Legal services pays a median of $59,800, and 74 percent of paralegals work in legal services, so that's the figure describing a law firm.
The national band, May 2024:
| Percentile | Annual wage |
|---|---|
| 10th | $39,710 |
| 25th | $48,190 |
| 50th (median) | $61,010 |
| 75th | $78,280 |
| 90th | $98,990 |
Twenty fifth to seventy fifth is about $30,000 of room, most of a second salary. Where you sit in that band is a real decision and no survey is going to make it for you.
For intake and case management you're building the benchmark yourself, since there isn't one to buy. Track what you pay and how long each person lasts in the seat. Three years of your own numbers is worth more than any guide, because it describes your caseload and your market instead of a national average of a job nobody has defined.
Nobody can tell you what a plaintiff firm pays
The two BLS programs disagree about how many paralegals exist, 376,200 in the projections against 367,220 in the wage file, because they're built differently and treat self employment differently. Growth figures here come from the projections and pay figures from the wage file, which is what that difference calls for, but the two are not interchangeable inside one calculation. A projection also isn't a measurement, and if generative AI reshapes document review faster than the model assumed then the 2024 to 2034 numbers will look silly in hindsight. The biggest limit is the one this whole thing is about: none of it cuts plaintiff against defense, because BLS classifies by industry and occupation and has never coded a firm by how it gets paid, so $59,800 has the insurance defense shop and the contingency practice down the road sitting in the same average. I can't tell you what a plaintiff firm pays. Nobody can, including the people selling you benchmarks.
It's two files if you want to redo it. The openings and separation rates are in occupation.xlsx from the employment projections, Table 1.10, row 23-2011. The wages are in oesm24nat.zip, the May 2024 national OEWS file, filtered to the same code.
What it costs you
A tenth of your operation leaves every year at minimum, and it never appears as a line item. It appears as three months of everybody else covering, which is why most partners have never priced it.
If the work that seat does lives entirely in one person's head and habits, you rebuild it from scratch on that cycle, forever, and the rebuild is the expensive part rather than the recruiting. Work that follows rules should sit somewhere that doesn't resign. That's most of what we do.
Whether to fill the seat again at all is a separate question, and we wrote about that one already.