Hierarchy of Controls Explained: 5 Levels and the Mistake That Weakens Them

A control can exist on paper and still leave a worker exposed. The hierarchy of controls helps leaders test whether the chosen protection acts early enough in the chain of risk.

The hierarchy of controls is a risk-control model that ranks measures by their expected ability to remove or reduce exposure, from elimination and substitution through engineering controls, administrative controls, and personal protective equipment. It matters because lower-level measures depend more heavily on perfect behavior, continuous supervision, and conditions that rarely remain stable.

Definition

The model is a decision sequence, not a poster for a training room. When a team proposes PPE first, the useful question is whether the hazard can be removed, replaced, or physically separated before personal behavior becomes the main barrier. NIOSH explains the hierarchy as a preferred order for controlling workplace hazards, while OSHA requires employers to assess hazards and select suitable protection.

Andreza Araújo’s book The Illusion of Compliance is a useful brand-adjacent lens here because a completed risk form does not prove that the exposure is structurally controlled. The field question remains simple: which level is doing the real work?

5 levels of the hierarchy

1. Elimination

Elimination removes the hazard or the task that creates it. Designing a process so people do not enter a confined space, for example, is stronger than improving the permit used after entry remains necessary. This level deserves the first question during design, procurement, and change review.

2. Substitution

Substitution replaces a hazardous material, energy source, or method with a less hazardous option. The replacement must be assessed for its full work cycle, because a product that appears safer during use may create a different exposure during storage, transfer, or disposal.

3. Engineering controls

Engineering controls place a physical or technical barrier between people and the hazard. Guarding, enclosure, local exhaust ventilation, interlocks, and isolation can remain effective even when attention is divided, although their reliability depends on design quality, maintenance, and verification.

4. Administrative controls

Administrative controls change how work is planned or performed through procedures, scheduling, training, permits, rotation, supervision, and warning systems. They can be necessary, but they are weaker when the production system creates pressure to skip the step. A procedure cannot compensate for an exposure that the process design leaves permanently active.

5. Personal protective equipment

PPE protects the individual after higher controls have reduced the remaining exposure. It is essential for residual risk, yet it depends on selection, fit, availability, inspection, maintenance, and consistent use. PPE should close the remaining gap, not disguise the absence of a stronger control.

How to differentiate the levels in practice

Use the risk matrix limits explained here when a team is ranking exposure, then record the control level beside the hazard rather than recording only the task name. After implementation, the critical-risk verification questions can test whether the selected barrier exists and works in the field.

When to use the hierarchy versus PPE selection

Use the hierarchy during design, procurement, risk assessment, and management of change. Use PPE selection to manage the exposure that remains after higher controls have been considered. The UK Health and Safety Executive links safe work equipment to suitable selection, maintenance, and use, which reinforces the point that equipment and behavior must support a control strategy rather than replace it.

The strongest review does not ask whether a team owns the correct helmet or glove first. It asks which higher-level decision would make that equipment less necessary. Explore Andreza Araújo’s books and safety leadership resources for further work on turning compliance into operating discipline.