ASQ CRE: The 10 Most Common Mistakes

By Sandeep Bandyopadhyay, Senior aerospace engineer · PMP, MBA, Six Sigma

TL;DR

Most ASQ CRE marks are lost to habits, not to missing theory. The exam is open book, with 165 questions delivered and 150 scored, so the traps are loose references, a handbook written for an older Body of Knowledge, a percentage chased when the pass standard is a scaled 550 out of 750, and models chosen for the wrong situation. This post covers ten, in the order you meet them, from preparation to the engineering questions.

Original questions, not exam dumps. Every CertCrafter question is written from the published body of knowledge and the referenced codes. None is copied from a live exam, and CertCrafter is independent: not affiliated with or endorsed by any awarding body.

A technician preparing a car for crash and endurance testing in a laboratory
Photo: Testing the Tesla at Argonne National Laboratory (4) by Argonne National Laboratory (Wikimedia Commons), CC BY-SA 2.0, via Wikimedia Commons.

Most CRE candidates know enough reliability theory to pass. They lose marks by preparing for the wrong exam: the wrong handbook, a loose binder, a percentage that does not exist, a model that does not fit the data.

Video: ASQ CRE — Six weeks to CRE, one hour a night. ASQ CRE — Six weeks to CRE, one hour a night. Watch on YouTube · CertCrafter
Image: video thumbnail, source: CertCrafter on YouTube (@certcrafters).

The channel has a short on six weeks to CRE, one hour a night. The mistakes below are what that plan has to avoid.

Which preparation mistakes cost the most marks?

Four of the ten happen before you read a single question. They are cheap to fix and expensive to ignore.

Mistake 1: a binder that is not bound. All ASQ exams are open book, and reference material must be bound and stay bound. ASQ’s page names ring binders, spiral binders and screw posts as acceptable, and says hand-stapled documents not securely fastened in their covers are not allowed. The Test Center Administrator inspects your references before you enter. Loose printouts in a folder are a risk you do not need.

Mistake 2: the wrong handbook and the wrong Body of Knowledge. The 2025 Body of Knowledge governs the current exam and has five areas: Reliability Fundamentals, Risk Management, Probability and Statistics for Reliability, Reliability Planning, Testing and Modeling, and Lifecycle Reliability. Much of the market still teaches an older structure. ASQ’s own question bank is written to the 2009 Body of Knowledge with seven domains, and ASQ’s reference list still points at the 3rd edition of the CRE Handbook, which was written to the 2018 outline. The 4th edition, by Hulting and McShane-Vaughn, was published in 2025. Check the edition on your desk against the outline you will be examined on.

Mistake 3: chasing a percentage. Forum posts quote a number. ASQ does not. The standard is a scaled 550 out of 750, set for each form by a modified Angoff procedure, and the raw score that equals 550 changes with the form. Study every area to a working level instead of banking on one score.

Mistake 4: the wrong calculator. Computer-based exams give you an on-screen basic scientific calculator, and the test center can hand you a basic hand-held one on request. A programmable or graphing model is out, and the TI-89 is named. Practice the whole calculation on a basic one, including the exponentials.

How should you pace a 165-question open-book exam?

Mistake 5: treating open book as easy. The computer-based exam delivers 165 questions, of which 150 are scored and 15 are unscored and not identified. Exam time is 4 hours and 18 minutes. That is under two minutes a question. You will not look up much in that time, so tab the binder by topic and use it to confirm a method you already chose.

Exam format from ASQ's page: 165 questions delivered, 150 scored, 15 unscored, 4 hours 18 minutes, open book, scaled pass mark 550 of 750
The computer-based CRE in numbers. Illustration: CertCrafter, from Certified Reliability Engineer (CRE) Certification | ASQ.

On a plant, this is the same habit as a good reliability review. Nobody derives a Weibull fit from scratch in the meeting. They know which tool the situation calls for and have the reference at hand for the detail.

What do candidates get wrong on the reliability maths?

Mistake 6: reading MTBF as a median life. With a constant failure rate, reliability at the MTBF is exp(-1), about 0.368, not 0.5. By the time a population reaches its MTBF, most of it has failed. A maintenance planner who reads MTBF as “half my units last this long” will promise a replacement interval the fleet cannot meet. Our reliability fundamentals post works the case through.

Mistake 7: fitting a Weibull to a repairable unit. This one catches experienced people. A plant manager asks whether one gearbox is failing more often across three years of repair records. The reflex is to fit a Weibull to the times between failures and read beta. But the Weibull describes times to first failure across a population of non-repairable items. A repair process needs a trend view, and the direct answer is a cumulative plot of failure count against cumulative operating time: a straight line means a constant rate, and a bend shows the rate changing. One of our free questions is built on exactly this, and the wrong option is the Weibull fit.

Mistake 8: ignoring the suspensions. Real test data is rarely clean. Sixty units go on test, four fail and the other fifty-six are pulled or still running at mixed times. Throwing away the survivors discards most of the information you paid for. The question you will be asked is which estimation method uses each suspension at its own time and supports confidence bounds. Our other free question covers it, and the answer is maximum likelihood. In practice, the habit to build is to ask what the data looks like before reaching for a plot.

Worked example: a series string is worse than its parts

These numbers are invented for practice, so check the method and ignore the figures.

A string has ten parts in series. Each has a reliability of 0.99 over the mission. What is the reliability of the string?

  1. Pick the configuration. Every part must work, so reliabilities multiply.
  2. Multiply 0.99 by itself ten times.
  3. The product is about 0.904.
  4. State it in words: roughly nine chances in ten, which is worse than any single part.

The trap is answering 0.99, the reliability of one part, or averaging the ten. Wrong options in a question like this are built from those two slips.

The ASQ CRE practice test bundle works more of these under timed conditions, with every answer explained.

Ten blocks in a chain, each marked 0.99, leading to a string reliability of about 0.904
Ten good parts in series make a weaker string. Illustration: CertCrafter, from Certified Reliability Engineer (CRE) Certification | ASQ.
Solar panels on a test stand at a laboratory, an example of hardware under reliability test
Photo: Solar panel testing, Sandia National Laboratories by U.S. Department of Energy (Wikimedia Commons), Public Domain, via Wikimedia Commons.

Which availability figure is the question asking for?

Mistake 9: mixing the availability family. Inherent availability is MTBF divided by the sum of MTBF and MTTR, and it leaves out logistics delay and preventive maintenance. Achieved and operational availability count more of the downtime. A supplier quoting inherent availability for a unit your site loses to spare-part delays is quoting a number your operation will never see. In the exam, the first line of the question names the figure. Read that before you reach for any formula, because every wrong option is a correct calculation of the wrong availability.

What should you check before you apply?

Mistake 10: a thin application. ASQ asks for 8 years of on-the-job experience in one or more areas of the Body of Knowledge, and 3 of those years must be in a decision-making position, in a full-time, paid role. A degree waives part of the 8 years, and only one waiver can be claimed: a technical or trade diploma one year, an associate degree two, a bachelor’s degree four, a master’s or doctorate five. The exam fee shown on ASQ’s page is $550, a retake is $350, and members save $100 on the initial fee. Check the current page for dates and deadlines before you plan around them.

What to do next

Sort the ten by when they bite. Four are set up before you study: binder, handbook, percentage, calculator. One is pacing. Three are engineering judgment: MTBF, repairable units, censored data. One is a definitions habit and one is paperwork.

Start with the first four, since they cost nothing to fix. Then read the certification guide for the scoring and the Body of Knowledge, try the free sample quiz to see how the stems are written, and use the guides hub for the posts that go area by area.

Tonight, write down the edition of your handbook and the Body of Knowledge it was written for. If they differ, that is your first fix.

Try these free questions

  1. Testing ends with four failures among 60 units, the other 56 suspended at mixed times, and the team needs Weibull parameters with confidence bounds. Which estimation method suits this data set?

    1. Rank regression on X, since regressing time on the ranked probability suits small failure counts best.
    2. Rank regression on Y, since the plotted points give a visual check on the fit that maximum likelihood cannot give.
    3. Least squares on the untransformed times, since it is unaffected by the number of suspensions present.
    4. Maximum likelihood, which uses each suspension time directly and gives the information for confidence bounds.
    Show the answer and explanation

    Answer: D. With only four failures among 60 units and 56 suspensions at mixed times, most of the information sits in the survivals, and the method has to use it properly. Maximum likelihood writes a density term for each failure and a survival term for each suspension, so every unit enters at its own time, and the likelihood surface also yields the Fisher information from which bounds on beta and eta are computed. Heavy mixed censoring together with a requirement for bounds settles the choice. Rank regression, whether time is regressed on the median rank or the rank on time, must first convert suspensions into adjusted ranks and then fit a line through four plotted points, and bounds drawn from that regression rest on assumptions four points barely support. Least squares on untransformed times ignores the Weibull form altogether. Maximum likelihood does carry a known small sample bias in beta, so the estimate should be reported with its bounds and a bias correction considered.

  2. A plant manager asks whether failures of one conveyor gearbox are arriving faster across three years of repair records on that single repairable unit. What plot answers the question directly?

    1. A Weibull probability plot of the times between successive failures, taking the fitted beta value as the indicator of trend.
    2. A cumulative plot of failure count against cumulative operating time, where a bending curve shows the rate changing.
    3. A histogram of the times between failures, which shows whether their distribution is skewed.
    4. A Pareto chart of failure modes on the gearbox, which ranks the contributors by their counts.
    Show the answer and explanation

    Answer: B. The question concerns one repairable unit and whether its failures are arriving faster with age, which is a question about trend in a sequence of events rather than about a life distribution. Plotting cumulative failure count against cumulative operating time answers it directly: a straight line means a constant rate, a curve bending upward means deterioration and a curve bending downward means improvement, and nothing has to be assumed about a distribution to read it. Fitting a Weibull to times between failures is the standard trap, because the Weibull model applies to times to first failure across a non-repairable population and its beta says nothing about trend in a repair process, so a deteriorating system can still yield times between failures that fit a Weibull with beta near 1. A histogram of times between failures throws away the order in which they occurred, and the order is the whole question. A Pareto of failure modes ranks contributors without addressing rate.

Take the full free 10-question sample with scoring →

Frequently asked questions

What is the most common mistake on the ASQ CRE exam?

Choosing the wrong method for the situation. The exam is open book, so a formula is easy to find. Deciding whether the data is a repairable unit or a population of parts, or whether a figure is inherent or operational availability, is the skill being tested.

Can I bring a programmable calculator to the CRE exam?

No. ASQ's page says any silent, hand-held, battery-operated calculator with no programmable memory is allowed, and that calculators such as the TI-89 are not. A basic scientific calculator is also on screen in computer-based exams.

Is there a percentage I need to pass the CRE?

No percentage is published. The pass standard is a scaled 550 out of 750 set by a modified Angoff procedure, and the raw number of correct answers needed varies by form.

Official sources

  1. Certified Reliability Engineer (CRE) Certification | ASQ — accessed 27 October 2026

Exam rules change. Check the awarding body's official page before you apply. CertCrafter is independent and is not affiliated with or endorsed by any awarding body.

Was this helpful?

Comments

    Comments are read by a person before they appear. How we moderate.