Search this question and you get a wall of confident numbers with nothing behind them. Here is the honest version. There is no published national typing speed standard for medical transcriptionists or medical scribes. What is published, and what employers actually score you against, is accuracy: the Association for Healthcare Documentation Integrity recommends a quality assurance score of 98.0 as the minimum industry standard for documentation created by a healthcare documentation specialist or a scribe. Speed is a budget problem underneath that. Work the arithmetic of audio length against typing throughput and the practical floor lands around 60 to 65 net WPM, with returns flattening hard past about 90.

That is the whole post in a paragraph, but the arithmetic is worth seeing, because it changes what you practice. Below is where the numbers come from, a worked example you can run against your own speed, and the part almost nobody explains: why "98% accuracy" on a typing test and a 98 QA score are not the same measurement, and not even close.

What WPM do medical transcription jobs require?

Start with what is actually documented. The Department of Labor's occupational profile for medical transcriptionists on O*NET OnLine describes the work in detail: converting dictated reports into written documents, editing for spelling, grammar and medical terminology, expanding jargon and abbreviations, and verifying questionable information with the provider. It lists the technology involved, including voice recognition software and EHR platforms. It reports a median wage of $19.43 per hour, about $40,410 a year, across roughly 43,900 positions, with employment projected to decline slightly over the 2024 to 2034 period and around 7,400 openings projected.

What it does not list, anywhere, is a typing speed requirement. Neither does the AHDI quality toolkit. The number is not a standard. It is a screening convenience that individual employers set, and it varies enough between a small clinic and a high-volume transcription service that quoting a single figure back at you would be guessing.

So instead of guessing, derive it. Transcription pay and transcription deadlines are both denominated in audio minutes, not in words. That gives you a real constraint you can solve.

The reference point for how long transcription takes comes from research practice rather than the transcription industry's own marketing. Utah State University's guide for oral history and qualitative fieldwork transcription puts it plainly: as a general rule, allow four hours of transcription time for every one hour of audio. That is clean audio, one speaker, no medical vocabulary, no formatting template, no QA pass. It is a floor, not a ceiling, and it is the number that makes the rest of the math work.

Why is 98% accuracy the real gatekeeper?

The AHDI Healthcare Documentation Quality Assessment and Management Best Practices toolkit, revised May 2025, is the closest thing this field has to a published standard, and it is worth reading rather than quoting secondhand. It recommends a QA score of 98.0 as the minimum industry standard for work produced by a healthcare documentation specialist or a scribe. Its sample score sheets are blunt about the cutoff: 98 and above passes, below 98 needs improvement.

The critical detail is how that score is computed. Errors are not counted, they are weighted by consequence:

Error class	Point value	Examples from the toolkit
Critical	-3	Wrong patient or encounter, wrong medication or dose (Zantac for Xanax, 15 for 50, mg for mcg), wrong lab value, terminology misuse that flips meaning (hypo for hyper, negative for positive), incorrect side or site, unapproved abbreviations, failure to flag
Noncritical	-1	Misspelled medication or name, incorrect word form, sound-alikes (gait for gate), punctuation that obscures meaning, protocol failure
Minor	-0.5 or -0.25	Capitalization, redundant text, ordinary typos with no effect on meaning
Educational feedback	0	Items worth teaching that carry no deduction

Two scoring methods appear in the toolkit. One subtracts error values from a per-report score of 100, and AHDI notes that under this method a single critical error fails the document outright, regardless of length, because a 3 point deduction lands below 98. The other divides total error value by lines reviewed, multiplies by 100 and subtracts from 100. Their own worked example: 0.75 in total error value across 206 lines reviewed scores 99.6.

Run that second formula backwards and the bar gets vivid. To hold 98 on a 200 line report you need total weighted error value at or below roughly 4 points, which is about four noncritical errors across the entire document. Now compare that to a typing test reporting 98% accuracy, which typically means two characters in every hundred were wrong, unweighted, with a transposed letter counting the same as a wrong dose. Across a full report that is a great many more defects than four, and some of them would be the 3 point kind. The two numbers share a digit and measure nothing alike.

This is the single most useful thing to understand before you practice. A 98% score on a general typing test is not evidence you can meet a 98 QA standard. It is roughly an order of magnitude short. If you want a figure that means something here, test accuracy deliberately rather than reading it off the corner of a speed result, and understand how WPM is actually calculated before you compare your number to anyone else's.

Medical scribe or transcriptionist: do the targets differ?

The quality bar does not. AHDI's toolkit applies the same 98 recommendation to documentation created by a healthcare documentation specialist or a scribe, and applies the same review structure to both, including 100% concurrent review of newly hired or inexperienced staff before their work reaches the medical record.

What differs is the clock, and that changes what typing speed has to buy you.

A transcriptionist works retrospectively from audio. Playback is under your control. You can slow it, rewind it, stop it while you look up a drug name. Your typing speed sets your total throughput per audio minute, which sets your pay and your turnaround, but it never causes you to miss content.

A scribe works concurrently, in the room or on a live feed, at the pace of a conversation nobody is adjusting for you. There is no rewind. Speed here buys headroom: the gap between how fast you produce text and how fast the encounter produces content is the margin you use for listening, reasoning about what belongs in the note, and navigating the EHR. Fall below that margin and you are not slow, you are lossy.

That is why the same WPM number feels different in the two roles. For a transcriptionist, speed is an economic variable. For a scribe, below a certain point it becomes an accuracy variable, and accuracy is the thing being scored.

How does audio speed change the math?

Here is the worked example. Take one minute of dictation and count the words in it. Do this with your own source audio rather than trusting an average, because dictation pace varies enormously between providers, and pauses, spelled-out drug names and dictated punctuation all pull the word count around. For the table below I have used 140 words per audio minute as a working figure. Swap in whatever you measured.

Pure typing time for one audio minute is words divided by your net WPM. The workload for that audio minute, using the four-to-one reference above, is four minutes. So:

Your net typing speed	Typing minutes per audio minute	Share of the 4 minute workload spent typing	Minutes left for everything else
40 WPM	3.5	88%	0.5
50 WPM	2.8	70%	1.2
65 WPM	2.2	54%	1.8
80 WPM	1.8	44%	2.2
100 WPM	1.4	35%	2.6
120 WPM	1.2	29%	2.8

Read the last column, not the first. "Everything else" is the actual job: rewinding a mumbled passage, confirming a dosage, applying the account's formatting template, flagging a discrepancy, proofreading. At 40 WPM there is half a minute of it available per audio minute, which is not enough to do the work the QA sheet grades you on. At 65 WPM there is nearly two minutes. At 120 WPM there is 2.8 minutes, which is better, but only 1 minute better than 65 WPM was, for nearly double the speed.

Now scale it to a real note. A 4 minute clinic dictation at 140 words per audio minute is about 560 words, with a 16 minute workload budget:

Your net typing speed	Typing time for the note	Time left in the 16 minute budget
45 WPM	12.4 min	3.6 min
70 WPM	8.0 min	8.0 min
95 WPM	5.9 min	10.1 min

That is the whole argument for the 60 to 65 floor, and against chasing 120. Below about 60 net WPM, typing consumes so much of the budget that the accuracy work gets squeezed, and accuracy is what the 98 standard measures. Somewhere past 90, the budget stops being typing-bound and starts being bound by audio handling and lookups. Which is exactly why experienced transcriptionists put money into a foot pedal, text expanders and a good headset rather than into another 20 WPM. Once you are past 65 WPM, roughly half the budget is not typing at all, and those tools attack that half.

One honest caveat on the trend line. O*NET projects employment in this occupation to decline slightly through 2034, and lists voice recognition software among the tools of the trade. The role is shifting toward editing machine output rather than typing from scratch, which raises the value of the proofreading half of the skill set and lowers the value of raw pace. That does not make speed irrelevant. It does make it a worse thing to be the best at.

How do you build speed on medical vocabulary?

Generic typing practice will not move your medical throughput much, because the bottleneck is not your fingers, it is recall. You slow down at unfamiliar terms, and you slow down disproportionately: an unrecognized drug name costs you a lookup, not a keystroke.

Three things that work, in order of effect:

Practice on the text you will actually type. Take a de-identified note in the specialty you work in, or the operative report template your account uses, and drill on that rather than on generic prose. You can paste your own passage into a custom typing test and get a real WPM and accuracy score on it. The gap between your general speed and your speed on that text is your vocabulary deficit, measured. It is usually the biggest single number you can move.
Build a text expander library before you build speed. Long, high-frequency terms are the ones costing you. Expanding a handful of specialty-specific strings does more for throughput than a month of drills, and it removes a whole class of misspelling errors from the QA sheet at the same time.
Drill accuracy at a controlled pace, then let speed follow. Typing faster than your error rate can support produces work that fails QA, and correcting a critical error costs far more than typing slowly would have. Our post on typing speed requirements by role covers the general mechanics of this, including why net WPM over a longer sample is the only figure worth reporting.

How do you prove your speed to an employer?

Three things happen in practice, and they call for different evidence.

The posting states a number and never tests it. Common. A documented result on your resume or application is enough to clear the screen, and the specifics matter: state net WPM, accuracy, test length and the date.

The employer tests you at interview, on their text. This is the norm in transcription and scribing, and it is usually medical text, sometimes their own template. Preparation is practicing on medical passages, not on quotes and pangrams. Our medical transcription typing test exists for exactly this rehearsal.

The employer requires a proctored or in-person certification. Some hospital systems, staffing agencies and government employers will only accept a test they administered or one from a designated proctored provider. Where that is the requirement, no online certificate qualifies, including ours. Say so early in the process and ask what they accept, rather than paying for a document that will be declined. What usually works there is asking to sit their test, or going through the proctored provider they name.

For the first two situations, a verifiable result is worth more than a screenshot, because a screenshot proves nothing to a hiring manager who has seen edited ones. TypeTimer's free result card gives you the number and is genuinely free, no account and no payment. The $4.99 certificate adds the part that a screenshot cannot: a public lookup page an employer can open themselves to confirm the score, the date and the test length. That is the whole difference, and it is worth being plain about it.

Frequently asked questions

Can I be a medical transcriptionist at 50 WPM?

You can do the work, but you will be working at a loss of margin. At 50 WPM, using the figures above, typing eats about 70% of a standard workload budget for a given audio minute, leaving under a minute and a half per audio minute for lookups, formatting, flagging and proofreading. Since pay is effectively per audio minute and quality is scored on the output, 50 WPM tends to mean either lower earnings or a squeezed QA margin. It is a fine place to start and a poor place to stay. Getting to 65 is a much bigger improvement than getting from 65 to 90.

What accuracy do hospitals require?

The published recommendation is AHDI's: a QA score of 98.0 as the minimum industry standard, with 98 and above passing on its sample score sheets, and one of the sample worksheets in the toolkit sets the bar at 98.5%. Individual facilities and transcription service organizations set their own contract terms, so check the specific one. Understand that this is a weighted score, not a character accuracy rate. A single critical error, worth 3 points, fails a document outright under the per-report scoring method.

Is medical transcription faster or harder than general typing?

Slower and harder, on the same fingers. Terminology forces lookups, dictation quality forces rewinds, and formatting templates add work that has nothing to do with typing. Most people find their medical WPM sits meaningfully below their general WPM until the vocabulary becomes automatic. Measuring both, on the same day, is the most useful diagnostic you can run.

Do employers test with medical text?

Usually, yes, for transcription and scribe roles specifically. Expect medical vocabulary, and often a passage close to the kind of note the role produces. A general typing test result is still worth having as evidence, but it is a poor rehearsal. Practicing on medical passages closes a gap that raw speed practice will not.

Measure the thing you will actually be tested on. Paste a de-identified note or a passage of medical text into the custom typing test, take it at a length that matches the employer's test, and see what your real medical WPM and accuracy are rather than your prose numbers. If the result is one you want on an application, a verified typing certificate records the speed and the accuracy together at $4.99, with a public lookup page an employer can check without taking your word for it.

Medical Transcription Typing Speed: Real Targets