Does a passing score of 700 out of 1000 mean I need 70 percent correct?

No. 700 out of 1000 is a scaled score, not a percentage of questions answered correctly. The scale is a fixed reporting range, and the raw number of questions you need right varies by exam form because of equating. On a harder form you can pass with fewer correct answers; on an easier form you may need more.

Why do two people pass the same exam with different numbers of correct answers?

Because exams come in multiple forms of slightly different difficulty. Vendors use equating so that a given scaled score always represents the same knowledge level, which means the raw number of correct answers required to reach the passing scaled score shifts between forms.

What are unscored or beta questions and do they count?

They are trial questions vendors seed into live exams to gather difficulty data. They do not count toward your score, but you are almost never told which ones they are, so you should answer every question as if it counts.

What is the passing score for Microsoft certification exams?

Microsoft role-based and fundamentals exams pass at 700 on a scale of roughly 1 to 1000. This is a scaled score, not 70 percent of questions correct.

What are the AWS and CompTIA passing scores?

AWS uses a 100 to 1000 scale: Cloud Practitioner passes at about 700, associate exams at about 720, and professional and specialty exams at about 750. CompTIA uses a 100 to 900 scale: Security+ passes at 750, Network+ at 720, and A+ requires Core 1 at about 675 and Core 2 at about 700.

Does Cisco publish a passing score for each exam?

No. Cisco reports scaled scores in a range that typically runs from roughly 300 to 1000, but it does not officially publish a fixed passing mark per exam, and the cutoff can vary. Figures you see online are unofficial estimates.

What score do I need to pass a Google Cloud exam?

Google Cloud exams are reported only as pass or fail with no numeric or scaled score. Community guidance often cites around 70 percent, but Google does not publish this figure, so treat it as an unofficial rule of thumb.

How long do I have to wait before retaking a failed exam?

Waiting periods are vendor-dependent, but a common pattern is a short wait (often around 14 to 24 hours) before a second attempt and a longer wait (frequently around 14 days) before later attempts, with a cap on attempts per rolling 12 months. Always confirm the official policy for your exam.

Should I aim to pass by exactly the minimum score?

No. Because weighting, unscored questions, and form difficulty are hidden from you, aim to score comfortably above the pass mark on realistic practice so normal exam-day variance does not push you below the line.

Certification Exam Passing Scores Explained (Scaled Scores and 700/1000)

You finish a certification exam, the screen shows a number like 720 or 700, and you have no idea whether that means you got 72 percent of the questions right. In most cases it does not. Nearly every major IT certification uses a scaled score, which is a deliberately abstract number designed to be fair across many different versions of the same exam. This article explains what scaled scores actually mean, why the number of questions you need to answer correctly moves around, and what the passing mark really is for each of the big vendors as of 2026.

Why a scaled score is not a raw percentage

A raw score is simple: it is the number of questions you answered correctly, or that number expressed as a percentage. A scaled score is a transformed number that maps your raw performance onto a fixed reporting range, such as 100 to 1000 or 200 to 800. The scale itself is arbitrary. A pass mark of 700 out of 1000 does not mean you needed 70 percent of the questions correct. It means your measured ability, based on the specific questions you saw, landed at or above the point the vendor defined as competent.

The reason vendors do this comes down to fairness. Exams exist in multiple forms, or versions, that draw from a large pool of questions. No two candidates necessarily see the same set, and some question sets are harder than others by chance. If everyone had to hit the same raw percentage, a candidate who happened to draw a tougher form would be penalized for something outside their control. Scaling corrects for this.

How scaling works and why the raw number needed shifts

Behind the scenes, vendors use a statistical process called equating. Every question in the bank is analyzed for difficulty. When an exam form is assembled, its overall difficulty is measured and the raw-to-scaled conversion is adjusted so that a given scaled score always represents the same level of knowledge, no matter which form you took.

The practical consequence is that the number of questions you must answer correctly to pass is not fixed. On an easier form, you might need to get more questions right to reach the passing scaled score, because each correct answer is worth less. On a harder form, fewer correct answers can get you to the same scaled score. This is why two people can pass the same certification having answered a different number of questions correctly, and why you should never assume a fixed percentage.

Weighted questions

Some exams weight questions differently. A complex, multi-step, or performance-based item may be worth more toward your scaled score than a simple single-answer multiple-choice question. CompTIA performance-based questions are a well-known example: these interactive simulations often carry more weight than standard items. The takeaway is that not every question contributes equally, so skipping a hard performance task to save time can cost you more than skipping an easy recall question.

Beta, unscored, and pilot questions

Many exams quietly include questions that do not count toward your score at all. These are variously called unscored, pilot, beta, or trial questions. Vendors seed them into live exams to gather difficulty data before deciding whether to score them on future forms. You are almost never told which questions are unscored, so you should answer every question as if it counts. This also means the number of questions that actually contribute to your score is smaller than the total on screen, which is another reason a naive percentage calculation will mislead you.

Practical rule: treat every question as scored, aim to finish with time to review, and do not try to reverse-engineer your percentage from the question count. You cannot see the weighting or the unscored items.

Vendor-by-vendor breakdown (as of 2026)

Passing marks and scales change over time, and some vendors deliberately do not publish a figure. The following reflects the situation as of 2026. Where a vendor does not disclose a number, we say so rather than inventing one. Always confirm against the official exam page before your test date.

Microsoft

Microsoft role-based and fundamentals exams (the AZ, AI, DP, MS, PL, SC, and MB series) are reported on a scale of roughly 1 to 1000, and the passing score is 700. This is the classic case where 700 out of 1000 is mistaken for 70 percent. It is a scaled score, not a percentage of questions answered correctly. Microsoft is explicit that the number of questions and their weighting vary, so you should not interpret 700 as needing 70 percent right.

AWS

AWS certification exams use a scaled score from 100 to 1000. The passing mark depends on the tier: the Cloud Practitioner foundational exam passes at about 700, associate-level exams (such as Solutions Architect Associate) pass at about 720, and the professional and specialty exams pass at about 750. AWS reports a pass or fail result plus your scaled score, and like most vendors it uses compensatory scoring, meaning strength in one domain can offset a weaker domain as long as your overall scaled score clears the bar.

CompTIA

CompTIA exams are scored on a scale of 100 to 900. Passing marks vary by exam: Security+ passes at 750, Network+ at 720, and the A+ certification requires passing two exams, with Core 1 at about 675 and Core 2 at about 700. CompTIA exams often include performance-based questions that carry heavier weight, so treat those simulations as high-value items rather than something to rush.

Cisco

Cisco reports exam results on a scaled range that typically runs from roughly 300 to 1000. Importantly, Cisco does not officially publish a fixed passing score for each exam, and the passing mark can vary from exam to exam and over time. You will often see figures like 825 or 850 cited online for specific exams, but these are unofficial estimates and Cisco does not confirm them. Plan to know the material thoroughly rather than aiming for a rumored cutoff.

Google Cloud

Google Cloud certification exams (such as Associate Cloud Engineer and the Professional exams) are reported simply as pass or fail. Google does not provide a numeric score, a scaled score, or a per-domain breakdown. Commonly cited community guidance suggests you need somewhere around 70 percent to pass, but Google does not publish this figure, so treat it as an unofficial rule of thumb only. Because you get no score detail, there is no partial-credit insight to learn from on a fail.

ISACA (CISA, CISM)

ISACA exams, including CISA (Certified Information Systems Auditor) and CISM (Certified Information Security Manager), use a scaled score from 200 to 800, and the passing mark is 450. This 450 out of 800 is a scaled score and does not correspond to any fixed percentage of questions correct. ISACA's scale and passing mark are consistent across its major exams.

(ISC)2 (CISSP, CCSP)

(ISC)2 exams, including CISSP and CCSP, use a 1000-point scaled scale with a passing score of 700. The CISSP in particular uses computerized adaptive testing (CAT) in many regions, where the difficulty of each question adapts to your performance and the exam can end early once it has enough evidence to score you confidently. Even with adaptive testing, the reported pass mark is 700 on the 1000-point scale, and it is not a raw percentage.

Practice with realistic exam questions

How to read a score report

When you finish, most vendors show a pass or fail result immediately, along with a scaled score (except where the vendor reports pass or fail only, like Google Cloud). Below the headline, you often get a domain or objective breakdown showing your relative performance in each topic area. Read this carefully:

The headline number is a scaled score, not a percentage of questions correct. Do not convert it to a percentage.
Domain breakdowns are usually shown as bars or relative bands (for example, above, at, or below the target), not exact percentages, so they indicate strengths and weaknesses rather than precise grades.
Most vendors use compensatory scoring, meaning a strong domain can offset a weaker one. A few exams require minimum performance per section, so check the exam rules.
If you fail, the domain breakdown is your best guide to what to study before a retake.

What a failed attempt means and retake waiting periods

A fail is not a permanent verdict; it is data. Use the domain breakdown to target your weakest areas rather than re-studying everything evenly. Almost every vendor enforces a waiting period between attempts to discourage guess-and-repeat behavior and to give you time to actually learn.

Waiting periods are vendor-dependent and change over time, so always confirm the current policy, but the general pattern looks like this:

A common rule is a short wait (often around 14 to 24 hours) before a second attempt.
After a failed retake, longer waits (frequently around 14 days) typically apply before each subsequent attempt.
Many vendors cap the number of attempts allowed within a rolling 12-month period.
You usually pay the full exam fee again for each attempt unless you bought a specific retake package or voucher.

These figures are general patterns, not a guarantee. Microsoft, AWS, CompTIA, Cisco, and the others each set their own retake rules, and they revise them periodically. Check the official policy for your exam before you book.

Why you should aim comfortably above the passing line

Because scaled scoring, weighting, and unscored questions all hide the true math from you, targeting the exact passing mark is risky. You cannot see which questions count, you cannot see the weighting, and you cannot know in advance whether your form is easier or harder. If you are only good enough to scrape a pass on a favorable form, an average form can push you below the line.

A better strategy is to consistently score well clear of the passing mark on realistic practice, ideally hitting the high end of the reporting range on your practice attempts, so that normal exam-day variance (nerves, an unfamiliar question style, a tougher form) does not sink you. Aiming for real mastery rather than the minimum also means the certification actually reflects skills you can use on the job, which is the entire point.

Study the official exam objectives, not just a question dump, so weighting surprises do not hurt you.
Take timed, full-length practice exams under realistic conditions to build stamina and pacing.
Track your practice scores by domain and drive up your weakest areas until you are comfortably above the pass line everywhere.
Only book the real exam once your practice scores are consistently clear of the passing mark, not just barely over it.

Understanding scaled scores removes a lot of exam-day anxiety. Once you know that 700 out of 1000 is a bar to clear rather than a percentage to hit, you can focus on what matters: knowing the material well enough that the scale, the weighting, and the unscored questions simply do not matter.

Start practicing now