It is the most-asked question in social work exam prep, and it has a real answer that ASWB publishes: between 66 and 78 correct out of 110 scored questions.
That is not a hedge or an estimate from a prep company. It is on ASWB's own exam scoring page, read there on 21 August 2026, in a sentence that also tells you why it cannot be more precise:
Because each exam has multiple forms, there is no single correct number of questions that must be answered correctly to pass.
Sit with the width of that range for a second, because almost nobody does the arithmetic out loud. Sixty-six out of 110 is 60%. Seventy-eight out of 110 is 71%. Depending on which version of the exam you are handed, the line you have to clear moves by nearly eleven percentage points.
That single fact should change how you set your target, and most study advice ignores it completely.
Why the number moves
ASWB publishes several versions of each exam at any one time, mainly for security — someone retaking gets a different form with different questions. Forms cannot be identical in difficulty, so ASWB adjusts for it. From the 08/2026 Examination Guidebook:
Because individual exam questions may have minor differences in difficulty, and because questions change from one exam administration to the next, statistical adjustments are made to account for the small differences in the overall difficulty of individual examinations. These adjustments result in slight increases and decreases in the number of questions that must be answered correctly to pass each version.
And the purpose, in ASWB's words: "no test-taker receives an advantage or disadvantage because of the version of exam received."
So a harder form has a lower bar and an easier form a higher one. You cannot know in advance which you will get, you will not be told which you got, and the difference is not trivial. This is the single most important thing to understand about ASWB scoring, and it is the reason no honest source will give you one number.
ASWB's own counsel put the same point more bluntly, in a December 2022 column explaining how the exams are built and defended: "each form may have a slightly different raw passing score."
Where the passing standard actually comes from
Here is the part that gets left out entirely, and it is the most human thing about the whole process.
On 30 and 31 January 2026, sixty-five licensed social workers from across the United States and Canada sat in a room in San Antonio, Texas, for a two-day standard setting workshop. They worked in five panels — one for each exam category — facilitated by psychometricians. Every one of them had to hold a valid license and be practicing. ASWB named all sixty-five publicly.
What they did there is the thing people imagine is done by an algorithm. From ASWB's description of the method:
A passing score study panel engages in an iterative process to determine the likelihood that a minimally competent social worker would answer each item correctly. After many rounds of analysis, the panel recommends the number of items that must be correctly answered to ensure that the test-taker possesses entry-level competence.
They reviewed the examinations. They rated the questions for how likely a minimally competent practitioner would be to answer each one correctly. They worked out, together, what minimum competence means. Then they arrived at a number.
That number is the passing standard for the exam you will sit — and it is a new number, because the standard is set for what ASWB calls the anchor examination, "created after the blueprints are developed." The blueprints changed on 3 August 2026. New blueprint, new anchor exam, new standard-setting panel, new standard.
Which has a practical consequence worth stating plainly: anyone telling you what they needed to score, based on an exam they sat before August, is describing a standard that no longer applies.
It is not a percentage, and it is not a curve
Two misconceptions do most of the damage here, and they pull in opposite directions.
It is not a percentage. ASWB's counsel is explicit that "a passing standard on a high-stakes licensure examination is not based on a percentage of correct responses." The 60% to 71% figures above are arithmetic done after the fact, not a target anyone set. Nobody decided you should get 70%; a panel decided which questions a minimally competent social worker would answer correctly, and the percentage fell out of that.
It is not a curve either. You are not competing with the other people testing that month. The same column is unambiguous:
Licensure examinations are criterion referenced, meaning that candidate results are based on the test-taker's demonstration of competence, not on performance by peers (e.g., on a curve).
Pass rates are an outcome of the standard, not an input to it. If fewer people pass this year, that does not mean the bar rose; it means fewer people cleared a bar that had already been set.
This is also why pass rates by category tell you something about cohorts and nothing about your own exam. And note what nobody has yet: there is no published pass rate for the new exam. Every rate in circulation — including the ones we publish — describes exams sat under the old blueprint, against the old standard. The first meaningful figures for the 2026 exams are some way off.
About the "scaled score of 0 to 200"
You will read, on a lot of pages, that ASWB reports a scaled score between 0 and 200 and that passing falls somewhere around 70 to 100.
We went looking for that in ASWB's own material. Here is exactly where we looked and what we found:
| Source | "scaled score" | "0 to 200" |
|---|---|---|
| 2026 Examination Guidebook (full document, 08/2026) | 0 | 0 |
| ASWB exam scoring page (read 21 August 2026) | 0 | 0 |
Zero occurrences. The word "scale" does turn up twice in the Guidebook — in "full-scale, 122-question" practice tests, and in a 1–5 self-rating column on a study checklist. Neither is a score scale. What ASWB says instead is that scores are "reported as pass/fail, with the number of questions needed to pass the exam and the number of questions answered correctly shown." Raw counts of questions. No scale, no conversion, no 200-point ceiling.
We are not going to tell you that nobody has ever seen a scaled score on an ASWB report, because we cannot prove a negative about documents we have not read. What we can tell you is that it appears in no current ASWB document we can find, and that the description ASWB does publish — questions needed, questions answered correctly — is not a scaled score at all. If you are budgeting revision against a 0–200 scale, you are working from a model of this exam that its own publisher does not describe.
Twelve questions that do not count
The exam is 122 questions. 110 are scored. 12 are pretest questions that do not count, and they are, in the Guidebook's words, "mixed together on the examination" — you cannot identify them.
Two things follow.
Answer everything. The Guidebook notes that "a question without an answer will always be counted as incorrect." There is no penalty structure to game.
Do not read too much into one strange question. A question that seems unfairly obscure, or unlike anything in your material, has a reasonable chance of being one of the twelve that is being trialed on you rather than scored. ASWB's whole reason for including them is that a question has to be "proven" on live candidates before it can become a scored item. Noticing a hard question and spiraling about it is expensive; the arithmetic says roughly one question in ten is not counting at all.
What your score report actually shows
You get your result on the testing computer before you leave, and the unofficial report reaches your ASWBCentral dashboard about 24 hours later.
The report gives you the number of questions needed to pass and the number you answered correctly. That first figure is worth understanding: it is the number for your form. It is the only place the 66-to-78 question is ever answered precisely, and it is answered after the fact, for you alone.
There is an asymmetry worth knowing about in advance. If you fail, you get more information than if you pass. A failing report adds a breakdown of how many you answered correctly and how many there were in each content area — the diagnostic that makes a second attempt smarter than the first. A pass gets you the headline and the license. If you do fail, what actually happens next covers what to do with that breakdown, and the waiver most people miss.
One footnote on that waiver, because it is scoring-adjacent: it exists if your score was "within 10 correct answers of the passing score" — that is what "within 10 points" means, in ASWB's own phrasing on its waiver page. California and North Dakota do not permit 90-day waivers at all, and nobody may use more than two in twelve months for the same exam category.
Your score travels
A small relief that people rarely know until they need it. From the Guidebook:
Your passing or failing score is valid in all states and provinces for the same examination category.
And from the scoring page: "All jurisdictions that use the exams recognize the same passing standard."
You do not retake a passed exam because you move or apply somewhere else. One exam, one standard, everywhere it is used. The state-by-state variation is in everything around the exam — supervised hours, titles, application fees, which exam your license requires — which is what the licensure pages are for.
So what should you actually aim for?
Here is the arithmetic nobody puts in front of you.
The top of ASWB's published range is 78 out of 110. That is 71%. If your plan is "get about 70% and I'll be fine", your plan is to land exactly on the fail line of a harder-than-average form, having done no better than the minimum on an easier one.
Aim clear of the top of the range, not the middle of it. Consistently scoring in the low 80s on realistic full-length practice gives you a margin that survives a hard form, a bad night's sleep, and the three questions where you misread the qualifier. Consistently scoring at 70% gives you a coin flip you cannot see the odds of in advance.
Three more things worth building into how you practice:
Practice under the real constraint. The form variation is invisible; the time is not. 122 questions in 240 minutes is a hair under two minutes each, and the format has a break you cannot come back from.
Track by content area, not overall. Your score report — if you need it — is broken down that way. And the weights are close to even under the new blueprint, 35/33/32 and 36/32/32 on Clinical, which cuts both ways: there is no small area you can safely neglect, and no single area worth cramming out of proportion to the others. Fix your weakest one.
Do not convert practice percentages into predictions. A percentage on a practice bank is a measure of that bank, not of an ASWB form that has been equated against an anchor exam. Use it as a trend line for yourself, not as a forecast.
What we built for this
You cannot know your form's passing number in advance. What you can do is stop guessing about your own readiness, which is the part actually under your control.
Our full-length mocks run the real thing's shape — 122 questions, timed, weighted to your exam category's content areas, none of them repeating across mocks — and every question carries a full rationale on every option, not just the one you picked. Afterwards you get a per-content-area breakdown that looks like the one ASWB would give you, except you get it before it matters rather than after.
You can test the writing before paying for any of it. Chapter 1, Ethical Principles and Responsibilities is free and needs no account — eight complete lessons with practice questions in the largest content area on your exam. And the free readiness quiz is weighted the way your exam is, which makes it a better first data point than a raw percentage from anywhere else.