What the cause tree is
The cause tree (root cause tree) is an accident investigation method that starts from the final fact (the injury or damage) and rebuilds, as a tree, every fact that was necessary for it to happen. The goal is to find what to change in the company, not to find someone to blame.
The method was developed in France and spread by the INRS. Its strength is the focus on verifiable facts: each box in the tree is something that happened or existed on site, confirmed by observation, records or testimony. From the tree you separate immediate causes from basic causes and define the action plan.
It complements the 5 Whys: the 5 Whys follow a single line of cause, while the tree shows that an accident almost always needs several conditions at the same time.
Visual map of the cause tree
The map shows this article's case, from the final fact to the basic causes and actions.
How to build the tree, step by step
- Secure the scene and record it. Isolate the area and photograph it; note the position of machines, devices and materials before anything changes. Care for the victim first.
- Collect facts. Interview the victim, witnesses and the supervisor; check procedures, work orders, maintenance records and the risk assessment. Do it soon after the event, on site.
- Write each fact as a short sentence with no opinion: "the pedal had no guard", not "lack of attention".
- Define the final fact and put it at the top or the right of the page.
- For each fact ask: "What was necessary for this to happen?". Each answer becomes a fact linked to the previous one.
- Identify the relation between facts (see the next section) and draw the links.
- Keep going until you reach facts the company can control: rules, routines, design, maintenance, training.
- Classify, propose actions and follow up until you confirm they work.
The three relations between facts
- Succession: fact A leads to fact B, and only A. A simple chain.
- Conjunction (AND): two or more facts must exist together to produce the next one. Remove any one of them and the accident is prevented.
- Disjunction (OR): independent facts, each of which alone produces the next. All must be addressed.
Validity test: if fact B occurred and fact A had not existed, would B still have occurred? If yes, A is not necessary and the link is wrong.
Worked example: finger crushed in a press
Fictional case, created to illustrate the method. A hydraulic press operator was removing a part stuck in the die by hand. At the same moment a parts bin fell onto the foot pedal and the press closed on his finger. The table lists the 12 facts found and how they link.
| No. | Fact | Linked to fact | Relation |
|---|---|---|---|
| 1 | Operator had a finger crushed in the press | Final fact | – |
| 2 | The finger was between the dies | 1 | AND (both are required) |
| 3 | The press started its cycle | 1 | AND |
| 4 | Operator was removing a stuck part by hand | 2 | Succession |
| 5 | Parts frequently stuck in the die | 4 | AND |
| 6 | No removal tool at the workstation | 4 | AND |
| 7 | Clearing procedure did not stop and lock out the machine | 4 | AND |
| 8 | Foot pedal pressed by a parts bin that fell onto it | 3 | AND |
| 9 | Foot pedal had no guard | 3 | AND |
| 10 | Die had no scheduled preventive maintenance | 5 | Succession |
| 11 | Pedal replaced without a risk assessment of the change | 9 | Succession |
| 12 | Operating procedure did not cover clearing jams | 6 and 7 | Succession |
Reading from top to bottom: the accident (1) required two conditions at once, the finger in the zone (2) and the cycle started (3). Eliminate either one and the injury does not happen. Each condition, in turn, has its own branch.
Immediate causes and basic causes
With the tree done, the facts split into levels. Stopping at immediate causes leads to the conclusion that "the operator made a mistake" and to a safety reminder. The gain is in the basic causes, which explain why the condition existed and would keep existing.
| Level | Facts in the example | Question that identifies it |
|---|---|---|
| Immediate cause | 2, 3, 4 and 8 | What was wrong at the time of the accident? |
| Intermediate fact | 5, 6, 7 and 9 | Why was it that way? |
| Basic cause (management) | 10, 11 and 12 | Which company rule, routine or decision let this through? |
In the example, the basic causes are three management decisions: no preventive maintenance of the die (10), a pedal changed without a risk assessment (11) and a procedure with a gap on clearing jams (12). Without addressing them, another operator would fall into the same trap on another press.
From diagnosis to action plan
Each branch of the tree needs at least one action that cuts it. Prioritize by the hierarchy of controls: eliminate the hazard, guard by engineering, then procedure and, last, PPE and training. Record each action in a 5W2H plan with an owner and a due date.
| Action | Facts addressed | Control type | Owner (example) | Due |
|---|---|---|---|---|
| Install a guard on the foot pedal | 8 and 9 | Engineering | Maintenance | 15 days |
| Fit the press zone with guarding that meets the applicable machine-safety rules | 2 and 3 | Engineering | Safety and Engineering | 45 days |
| Supply a removal tool and fix it at the station | 4 and 6 | Engineering / administrative | Production supervisor | 7 days |
| Write a clearing procedure with stop and lockout | 7 and 12 | Administrative | Safety and Production | 20 days |
| Add the die to the preventive maintenance plan | 5 and 10 | Administrative | Maintenance | 30 days |
| Require a risk assessment for every change to an actuation device | 11 | Administrative (management of change) | Safety | 30 days |
See how this reasoning connects with other tools: the risk matrix and preliminary risk analysis should be reviewed with what the accident revealed; energy isolation follows LOTO; machine guarding is the subject of machine safety rules. Formal handling of the event follows the corrective action flow, and ISO 45001, in requirements and implementation, requires investigating incidents and acting on causes.
After the actions, track results over time. The main outcome indicator is the injury frequency and severity rate, but also include prevention indicators such as near misses reported and actions closed on time.
Common investigation mistakes
- Looking for someone to blame. If the conclusion is "unsafe act by the operator", the investigation stopped early. Ask why the act was possible and likely.
- Mixing fact with opinion. "Lack of attention" is not verifiable; "the pedal had no guard" is.
- Investigating late. After the area is cleaned and details are forgotten, the tree fills with assumptions.
- Stopping at immediate causes. The action becomes training or a warning and the risk remains.
- Administrative-only actions. Procedures and posters alone fail; look for an engineering solution first.
- Not verifying effectiveness. Define how and when you will confirm the branch was really cut.
- Ignoring the near miss. It has the same tree without the injury. Investigate it too.
Frequently asked questions
What is the cause tree method?
It is a method that rebuilds, as a tree, the facts needed for an accident to occur, starting from the final fact.
What is the difference between the cause tree and the 5 Whys?
The 5 Whys follow a single line of causes. The tree shows several simultaneous conditions and their AND and OR links.
What are immediate and basic causes?
The immediate cause is what was wrong at the time of the accident. The basic cause is the management, rule or design failure that allowed that condition.
Who should take part in the investigation?
Occupational safety, the area supervisor, a workers' representative, maintenance when equipment is involved and, when possible, the injured person.
When does the investigation end?
When every branch of the tree has an action with an owner and a due date, and effectiveness has been verified after implementation.
Does the cause tree work for near misses?
Yes. A near miss has the same structure of facts, and investigating it lets you act before anyone is hurt.
Sources
- ISO 45001:2018. Occupational health and safety management systems: requirements with guidance for use.
- ABNT NBR 14280. Cadastro de acidente do trabalho: procedimento e classificação.
- REASON, J. Human Error. Cambridge University Press, 1990.
- HOLLNAGEL, E. Barriers and Accident Prevention. Ashgate, 2004.
- HEINRICH, H. W. Industrial Accident Prevention. McGraw-Hill, 1931.
Keep studying: trusted sources and links
- ISO: ISO 45001, Occupational health and safety
- Ministério do Trabalho e Emprego: Normas Regulamentadoras (NRs) (in Portuguese)
Save this address: Manufatura Enxuta is your school for lean manufacturing. Come back whenever you need to study, with a complete article, a visual map and trusted sources. Follow one of the learning paths or check the A–Z glossary.
Want the tools ready to use?
Download the free "Problem Solving Kit" e-book, with PDCA, A3, 5 Whys and Ishikawa.
Download the e-book
About the author
Lean Manufacturing & Behavioral Management Specialist
Over 20 years in the automotive and metalworking industries (GM and Dana), Lean Manufacturing practitioner since 2006. SENAI instructor and mentor in Brazil’s Brasil Mais Produtivo program, delivering consulting, training and audits for 50+ companies, combining quality, productivity and people development.
