This study guide for BCSP's Certified Safety Professional (CSP) exam follows the board's current blueprint, CSP11 (Aug 1, 2025), domain by section. Each section gives the domain's weight, the tasks BCSP lists under it, the pairs of concepts that are easiest to confuse at a manager's level, and two worked items in the quiz at the end. [1][2]
BCSP publishes a percentage for each domain and nothing finer: no item counts, and no weights for individual tasks. The figures given as "about N of the 200 questions" below are our arithmetic from those percentages, so a given form can run a few items either way.
Two domains, Advanced Application of Safety Principles and Program Management, make up half the exam between them. The five smaller ones still decide results, because BCSP scores one combined total rather than a pass per domain.
Domain weights from BCSP's CSP11 blueprint, with our estimate of items out of 200.
D125%D225%D315%D49%D56%D610%D710%
0%Domains 1 + 2 = 50%100% = 200 questions
CSP11: each brick is one domain, as wide as its share of the exam. “About” is that share applied to the 200 questions — our estimate; BCSP prints percentages only. Changes index 2025CSP11 blueprint
About 50 of the 200 questions. The widest domain on the blueprint takes in prevention through design; process safety, from pressure relief and chemical compatibility to materials of construction and process flow diagrams; the everyday hazards of electrical work, falls, confined spaces, energy isolation, excavation, work over water and struck-by or caught-in exposure; facility life safety; fleet safety; materials handling from forklifts and aerial lifts to rigging, hoists and drones; and tools, ladders, hydraulics and robots.
BCSP words these tasks as applying principles, so the useful preparation is knowing why a control works and where it stops working, not only which rule requires it. Prevention through design is the thread: a hazard removed on the drawing board needs no permit, training or inspection later.
About 50 of the 200 questions. This is the manager's domain: benchmarking and gap analysis, performance standards and action plans, measuring culture, investigations with root causes and corrective action, management of change, system safety techniques, performance indicators, management and audit systems, written plans and policies, record retention, budgets and cost-benefit work, leadership theory, project tools such as RACI charts, and reading exposure data with descriptive statistics.
Several of these topics come as families of similar-sounding tools, each built for a different job; the three boxes below separate the members of three such families.
About 30 of the 200 questions. BCSP lists the whole risk process, from identification through analysis, evaluation and treatment to communication; job and process hazard analysis; the hierarchy of controls; the financial treatment of risk; and ranking and monitoring risks for emergencies, fire, health exposures and environmental compliance.
For the financial side, learn the treatments by what happens to the loss: avoidance drops the activity, loss prevention and reduction change the odds or the size of the loss, sharing and transfer move its cost to another party, and retention keeps it on your own books.
About 18 of the 200 questions. The tasks cover building and maintaining an emergency response plan for fire, severe weather, nuclear incidents, natural disasters, terrorism, chemical spills, utility failures and cyber events; incident command, business continuity and contingency planning; fire prevention, protection and suppression systems; moving and securing hazardous materials; and preventing workplace violence.
Three terms are easy to blur. The emergency response plan covers the first minutes and hours, the incident command system is the organizational structure that runs the response, and the business continuity plan sustains essential operations, or restores them, once people are safe. Each has its own owner, trigger and test.
About 12 of the 200 questions, the smallest share on the blueprint. BCSP lists pollution prevention and spill containment; hazardous materials under the GHS classification system, with storage and hazardous waste disposal; universal waste, recycling, labeling and remediation; sustainability in the supply chain; and aging infrastructure, asbestos, air pollution, climate change and ESG reporting.
The ordering of pollution prevention is worth fixing in memory: reduce at the source first, then reuse or recycle, then treat, with disposal as the last resort. Universal waste rules are a lighter-touch track for common hazardous wastes such as batteries, mercury lamps and some pesticides, not an exemption from hazardous waste law.
About 20 of the 200 questions. The domain follows the industrial hygiene cycle of anticipating, recognizing, evaluating and controlling exposures, across chemicals, radiation, noise, biological agents, heat and cold, indoor air, ventilation, combustible dust, silica, hot work and lasers. It adds epidemiology and public-health statistics, toxicology for exposure control, ergonomics and human factors, chemistry for containment volumes and the physics of energy, forces and stresses.
About 20 of the 200 questions. BCSP's tasks run the full cycle: assessing training needs and required qualifications, designing programs and materials for different learners, improving programs over time, judging effectiveness through surveys, observation, feedback and assessments, choosing delivery from classroom and simulation to computer-based, AI-assisted and coaching formats, and applying adult learning principles.
The classic frame for evaluation is Kirkpatrick's four levels: reaction (did trainees value it), learning (can they show the knowledge or skill), behavior (does the skill show up in daily work) and results (did the organization's outcomes change). A satisfied class proves only the first level, and each higher level costs more to measure.
For background reading, BCSP keeps a list of the books its item writers leaned on most; the table names one and links the rest. BCSP also sells a self-paced course and a half-length self-assessment; neither is required to apply. Both are listed in the table with their sources, and this guide draws nothing from them.
Fact
Detail
Source
Blueprint now used
The CSP11 blueprint, used for every candidate since August 1, 2025.
BCSP has announced no newer edition; its 2026 changes index lists no update to this exam. BCSP posted its last two changes indexes on January 6, 2025 and January 5, 2026.
CSP10 (V.2019.04.25), which CSP11 replaced on August 1, 2025, spread the exam over nine domains, the first of them a stand-alone Advanced Sciences and Math domain (9.95%). CSP10’s other domains: Management Systems 13.34%, Risk Management 14.49%, Advanced Application of Key Safety Concepts 14.69%, Emergency Preparedness, Fire Prevention, and Security 10.59%, Occupational Health and Ergonomics 12.05%, Environmental Management Systems 7.38%, Training/Education 10.18%, Law and Ethics 7.33%. CSP11 folds math into tasks such as data analysis, containment-volume chemistry and physics, and puts half the exam into two domains.
BCSP posts a list of books its item writers often consulted, such as the Accident Prevention Manual from the National Safety Council (14th edition). BCSP says each item rests on a published source, yet warns that the list is incomplete and no assurance of a pass.
BCSP’s paid CSP examCORE course (online, self-paced, pre- and post-tests, over 1,000 questions) and a half-length online self-assessment. BCSP keeps the examCORE authors apart from its exam-development panels and says finishing the course is no promise of passing.
From BCSP's CSP11 blueprint (V.2024.04.24), the archived CSP10 blueprint, BCSP's 2025 and 2026 changes indexes, its August 2025 reference list and its credential page; checked November 20, 2026.
Worked items, two per domain, with rationales
Answered 0 / 14 · matching the key 0
Each lettered option unfolds to its own reasoning, kept apart from the other three.
Domain 1 · Safety principles · 25%Item 1 · 14 in the set
An operator clears a paper jam every 15 minutes by reaching past a fixed guard; there is no interlock and no jam-clearing tool. Does this task qualify for the minor-servicing exception to OSHA's lockout/tagout standard?
AYes, because each jam takes under a minute
Why A is wrong. The exception has no time threshold; a short task with a hand in the danger zone still needs lockout unless other effective protection is present.
BNo, unless a supervisor watches each clearing
Why B is wrong. Watching does nothing to stop the machine moving while a hand is inside, so supervision cannot stand in for the protective measures the exception requires.
CYes, because it is routine and repetitive
Why C is wrong. Being routine, repetitive and integral to production is only the first condition; the exception also demands effective alternative protection, which is absent.
DNo, it lacks effective alternative protection
Why D is right. OSHA 1910.147(a)(2)(ii) excuses minor servicing only when alternative measures give effective protection; reaching past a fixed guard with no interlock or jam-clearing tool provides none.
Domain 1 · Safety principles · 25%Item 2 · 14 in the set
A process engineer must size emergency relief for a foamy, exothermic polymerization that can run away. Which data source is MOST appropriate for the sizing basis?
ADSC screening scans of each pure reactant
Why A is wrong. DSC scans of single reactants flag hazards but miss mixture interactions and give no pressure data, so they are a screening step, not a sizing basis.
BThe literature heat of reaction at normal conditions
Why B is wrong. A heat of reaction at normal conditions gives total energy but not the temperature and pressure rise rates during a runaway, which drive relief sizing.
CAdiabatic calorimetry of the actual mixture
Why C is right. Low-thermal-inertia adiabatic tests on the real reacting mixture reproduce runaway conditions and give the rates of temperature and pressure rise, including foamy two-phase behavior, that the relief design needs.
DThe relief valve vendor's rated capacity chart
Why D is wrong. A vendor chart tells what a valve can pass once the required relief rate is known; it cannot tell how much the runaway must vent.
Domain 2 · Program management · 25%Item 3 · 14 in the set
A log-normal probability plot of 15 air samples is nearly straight (correlation coefficient 0.98), with a geometric mean of 24.5 ppm and a GSD of 1.5. Which conclusion is MOST appropriate?
ANormal; report the mean and SD
Why A is wrong. The straight line appears on a log scale, which shows normality of the logarithms, not of the raw concentrations.
BSkewed; report the mean and SD
Why B is wrong. Air-sampling data are indeed right-skewed, which is exactly why the arithmetic mean and SD describe them poorly; the log-normal summary is used instead.
CLog-normal; report the GM and GSD
Why C is right. A straight line on log-probability paper (r = 0.98) shows the logarithms are close to normal, so the data are log-normal and are summarized by the geometric mean and geometric standard deviation.
DInconclusive; report GM without GSD
Why D is wrong. With r = 0.98 the fit is strong, not inconclusive, and a geometric mean without its GSD hides the spread that drives exceedance estimates.
Domain 2 · Program management · 25%Item 4 · 14 in the set
A vendor remotely pushes a firmware update to a facility's safety instrumented system (SIS) logic solver without a site management of change (MOC) review. What is the PRIMARY safety concern?
AThe trip functions may not act as validated
Why A is right. An SIS is validated to perform defined trip functions, and an unreviewed firmware change can alter trip logic or timing unnoticed, so the change needs MOC review and re-verification before it is trusted.
BThe update may void the vendor's service warranty
Why B is wrong. A warranty is a commercial matter; it does not tell anyone whether the trips still work.
CSite records will show the wrong version
Why C is wrong. An out-of-date record is a documentation gap; the danger is that unreviewed code is now running the safety functions.
DOperators may not be trained on new HMI screens
Why D is wrong. A logic solver firmware update does not normally change the operator screens, and retraining would not address untested trip logic.
Domain 3 · Risk management · 15%Item 5 · 14 in the set
A board asks for the metrics that give the BEST view of the company's exposure to fatal and life-altering injuries. What should the safety professional report?
AWorkers' compensation costs and claim counts
Why A is wrong. Compensation costs and claim counts follow claim volume and medical pricing and arrive late; they do not single out serious-injury potential.
BSIF precursor events and critical-control checks
Why B is right. Fatal and life-altering exposure shows up in events with serious-injury potential and in whether critical controls such as lockout, fall protection and permits are verified as working.
CThe total recordable incident rate (TRIR) trend
Why C is wrong. A falling TRIR has repeatedly coexisted with fatalities, because the rate is driven by frequent minor cases, not by serious-injury potential.
DThe DART rate compared with the industry average
Why D is wrong. DART counts cases with days away or restricted work, mostly sprains and strains, so it can look good while fatal exposure is unchanged.
Domain 4 · Emergency management · 9%Item 6 · 14 in the setRestart
After hurricane damage, a facility is preparing to re-energize its process equipment. Which item must be on the go/no-go checklist before restart?
ALOTO clearance and functional interlock tests
Why A is right. Restart after storm damage needs proof that locks and tags have been formally cleared and that the safety interlocks still work, because water and wind damage often hide in wiring and sensors.
BThe equipment vendor's certificate alone
Why B is wrong. A certificate issued before the hurricane cannot show that the safeguards survived the damage; on-site verification is still needed.
CPower available and production staff scheduled
Why C is wrong. Power and staffing show the plant can run, not that its safeguards will work when the equipment is energized.
DAn exterior visual check and production schedule
Why D is wrong. A walk-around from outside cannot reveal flooded wiring or failed sensors, and a production plan says nothing about safety.
Domain 4 · Emergency management · 9%Item 7 · 14 in the set
A chemical plant next to a residential area uses a toxic gas that could travel off site. Which warning configuration BEST reaches both workers inside the plant and residents outdoors?
AOutdoor sirens loud enough to cover the plant too
Why A is wrong. Louder outdoor sirens still leave workers in noisy buildings unable to hear them, so volume alone does not close the indoor gap.
BOutdoor sirens plus indoor visual and text alerts
Why B is right. Each group gets a channel that reaches it: sirens for residents outdoors, and visual and text notification for workers inside noisy buildings.
CIndoor PA announcements and a phone tree to officials
Why C is wrong. Indoor PA announcements cover people inside the plant, but a phone tree to officials adds a delay before anyone outdoors in the neighborhood is warned.
DOne high-output siren at the plant fence line
Why D is wrong. A single fence-line siren gives no indoor coverage and leaves parts of the community at the edge of its range.
Domain 5 · Environmental management · 6%Item 8 · 14 in the set
A plant is designing a groundwater monitoring network to detect any release from its underground solvent tanks. Which well layout is MOST appropriate?
ASeveral wells upgradient of the tank area
Why A is wrong. Wells placed only upgradient show background water quality; a plume moving away from the tanks never passes them.
BOne well upgradient, several downgradient
Why B is right. One well upgradient sets background quality and several downgradient wells sit in the path of groundwater flow, so a release shows up as a difference between them.
CWells placed evenly around the property line
Why C is wrong. Evenly spaced boundary wells ignore the flow direction and would detect a plume late, after it has crossed most of the site.
DOne well, drilled through the tank pit
Why D is wrong. A single well drilled through the tank pit risks opening a pathway for contamination and cannot show background quality or plume direction.
Domain 5 · Environmental management · 6%Item 9 · 14 in the set
A facility's stormwater results compared with benchmarks are: Outfall A copper 0.50 mg/L (benchmark 0.063); Outfall B zinc 0.25 mg/L (benchmark 0.12) and TSS 250 mg/L (benchmark 100); Outfall C copper 0.08 mg/L (benchmark 0.063). Which outfall should get corrective action FIRST based on exceedance severity?
AOutfall A
Why A is right. Severity is measured as a multiple of each benchmark: copper at A is 0.50 ÷ 0.063 ≈ 7.9 times its benchmark, far above B (about 2.1 and 2.5 times) and C (about 1.3 times).
Why B is wrong. B has two exceedances and the largest raw number, but TSS is reported on a different scale; relative to their benchmarks, both of B's results are smaller exceedances than the copper at A.
COutfall C
Why C is wrong. The copper at C is only about 1.3 times its benchmark, the smallest exceedance of the three outfalls.
DOutfalls A and B equally
Why D is wrong. Every exceedance needs follow-up, but equal treatment ignores that A's copper is far further over its benchmark than either result at B.
Domain 6 · Occupational health · 10%Item 10 · 14 in the setReverberation
A 10 m × 8 m × 4 m room has absorption coefficients of 0.80 for the ceiling, 0.05 for the floor and 0.15 for the walls. Using RT60 = 0.161 V ÷ A, where A = Σ(surface area × coefficient) in m², what is the reverberation time?
A0.18 s
Why A is wrong. This value comes from using the imperial constant 0.049 with metric dimensions; with meters the constant is 0.161.
B0.51 s
Why B is wrong. This result applies one unweighted mean coefficient to the whole surface; each surface has to be multiplied by its own coefficient.
C0.58 s
Why C is right. V = 320 m³ and A = 64 + 4 + 21.6 = 89.6 m² (ceiling, floor and four walls, each weighted by its coefficient), so RT60 = 0.161 × 320 ÷ 89.6 ≈ 0.58 s.
Room volume=10 × 8 × 4 = 320 m³
Total absorption=80 × 0.80 + 80 × 0.05 + 144 × 0.15 = 89.6 m²
Reverberation time=0.161 × 320 ÷ 89.6 =0.58 s
D0.76 s
Why D is wrong. Leaving the 144 m² of wall out of the absorption total shrinks A to 68 m² and inflates the reverberation time.
Domain 6 · Occupational health · 10%Item 11 · 14 in the setNanoparticles
A plant handling nanoscale titanium dioxide finds personal mass concentrations far below the OSHA PEL for TiO2 dust. Why may this NOT show that exposure is adequately controlled?
AMass sampling overstates nanoparticle exposure
Why A is wrong. Mass sampling tends to understate the dose for nanoparticles, because very little mass carries a very large surface area.
BNanoparticles enter mainly through the skin
Why B is wrong. Skin entry is not the main route for airborne nanoscale TiO2; inhalation is, so the weakness of the data lies in the metric, not the route.
CNanoparticles are too large to reach the alveoli
Why C is wrong. The opposite holds: particles at the nanoscale deposit efficiently deep in the lung, including the alveolar region.
DToxicity tracks surface area more than mass
Why D is right. Nanoscale particles carry a huge surface area and particle count at low mass, and their biological effect follows those measures, so a low mass result can still hide a meaningful dose; NIOSH recommends a lower limit for ultrafine than for fine TiO2 for this reason.
Domain 7 · Training · 10%Item 12 · 14 in the set
Some trainees apply a new job hazard analysis process very well, while others barely use it. The safety manager wants to learn what separates the two groups. Which evaluation approach is designed for this?
AInterviewing the best and worst adopters
Why A is right. Interviewing the best and worst adopters is the Success Case Method: studying the most and least successful trainees in depth shows what helped or blocked application, such as supervisor support or tools.
BAveraging all trainees' posttest scores
Why B is wrong. An average blends the two groups together and measures learning, not on-the-job use.
CA reaction survey and quiz at course end
Why C is wrong. A reaction survey and quiz at course end are collected before anyone tries the process on the job, so they cannot explain later differences in use.
DCounting completed JHAs for each crew
Why D is wrong. A count shows how much the process is used, not why some trainees apply it and others do not.
Domain 7 · Training · 10%Item 13 · 14 in the set
A forklift operator assessment scores candidates mainly on how quickly they move three pallets through a course. What is the critical validity concern with this assessment?
AIt uses too few pallets to be reliable
Why A is wrong. More pallets would make the timing more consistent, but a more reliable speed score still measures the wrong thing.
BIt ignores prior experience
Why B is wrong. The test should measure current safe performance; past experience is not the quantity being certified.
CIt is not timed the same way each run
Why C is wrong. Timing each run the same way improves reliability, not validity; the problem is that time is the scored behavior.
DIt rewards speed over safe practice
Why D is right. A valid assessment measures the behavior it is meant to certify; scoring by time rewards the rushing a forklift test should catch.
Domain 3 · Risk management · 15%Item 14 · 14 in the set
To reduce the worst-case release size, a plant replaces 1-ton chlorine containers with 150-lb cylinders. Which new risk should the change review MOST carefully consider?
AA larger worst-case release per event
Why A is wrong. The change does the opposite: a 150-lb cylinder holds far less than a 1-ton container, so the worst single release shrinks.
BFar more frequent hookups and disconnections
Why B is right. Supplying the same chlorine from small cylinders means far more frequent hookups, disconnections and deliveries, and each one is a chance for a leak; the change review has to weigh that against the smaller worst case.
CHigher toxicity and corrosivity of the chlorine
Why C is wrong. Toxicity and corrosivity are properties of chlorine itself and do not change with the package.
DHigher vapor pressure in smaller cylinders
Why D is wrong. Vapor pressure depends on the substance and its temperature, not on the size of the container.
Practice set, format and fees for this exam
Planning study around the current blueprint
Is BCSP about to replace the CSP11 blueprint?
Not so far. The next January changes index is where a successor would first appear, so it is the document to check before buying new material.