How to Do a REBA Assessment: Difference from RULA and Sample Scoring

How is a REBA assessment done? From trunk, neck, leg and arm angles to a 1–15 score step by step; RULA's 1–7 score, action levels and a floor-to-waist box lifting example.

Translated from the Turkish original · Türkçe aslı

A REBA assessment is an ergonomics method that observes the posture at the most demanding moment of a task, scores the trunk, neck, leg, upper arm, lower arm and wrist angles, and, with adjustments for load, coupling and activity, produces a single risk score between 1 and 15. A score of 1 means negligible risk, 2–3 low, 4–7 medium, 8–10 high, and 11 or above very high risk.

For fixed-workstation tasks where arm, wrist and neck loading dominate, the sister method RULA is used. Below we explain both methods step by step with the angle ranges and action levels from their original publications, and work through a hypothetical scoring of lifting a box from the floor; keep the official worksheet to hand for the intermediate tables.

What are REBA and RULA, and when is each used?

RULA (Rapid Upper Limb Assessment) was published by McAtamney and Corlett in 1993 in the journal Applied Ergonomics. It was developed to screen upper limb risk in tasks such as garment manufacturing, computer work and assembly; its final score ranges from 1 to 7. REBA (Rapid Entire Body Assessment) was published by Hignett and McAtamney in 2000 as a technical note in the same journal. It was designed for the unpredictable whole-body postures found in healthcare and service industries, was developed on more than 600 posture examples, and also factors load, coupling and activity into the score; its final score ranges from 1 to 15.

The choice depends on the nature of the task. If the worker sits or stands at a fixed workstation and the load builds up in the hand, wrist and shoulder, RULA gives a more detailed result. If the trunk bends, the knees flex, loads are lifted or the posture changes constantly, REBA is chosen. Both are screening methods; they do not diagnose, they prioritise. In Türkiye, the Regulation on Manual Handling Operations (Elle Taşıma İşleri Yönetmeliği, 24.07.2013, Official Gazette No. 28717) does not set a fixed weight limit; the Ministry's implementation guide also treats the assessment methods as recommendations.

Which posture is selected before starting the assessment?

Both methods score a single moment, so the choice of that moment determines the result. The worker is first observed over several cycles. The posture to be scored is selected as the one held longest in the cycle or the moment when the load is heaviest; if these differ, both are scored separately. The right and left sides are assessed separately, and if the load is in one arm, the loaded side is used. Angles are measured not by eye but on a photograph or video frame taken from the side; if the camera is not perpendicular to the body, the angle appears smaller than it is. The load's weight is determined by weighing it, and the number of repetitions with a stopwatch.

How is RULA scored step by step?

The first step is Group A: upper arm, lower arm and wrist. The upper arm scores 1 between 20° extension and 20° flexion, 2 in extension beyond 20° or in 20–45° flexion, 3 in 45–90° flexion and 4 above 90°; +1 is applied if the shoulder is raised, +1 if the arm is abducted, and −1 if the arm is supported or the person is leaning. The lower arm scores 1 in 60–100° flexion and 2 below 60° or above 100°; +1 if the arm crosses the body's midline or moves out to the side. The wrist scores 1 in a neutral position, 2 in 0–15° flexion or extension and 3 above 15°; +1 if there is radial or ulnar deviation. Wrist twist scores 1 in the mid-range and 2 near the end of the range. These four values are combined in Table A.

In the second step, muscle use and force are added to the Table A result. If the posture is mainly static (held for longer than 1 minute) or the movement is repeated more than 4 times per minute, +1 is added. The force score is 0 for an intermittent load up to 2 kg, 1 for an intermittent load of 2–10 kg, 2 for a static or repeated load of 2–10 kg or an intermittent load above 10 kg, and 3 for a static or repeated load above 10 kg or for sudden, shock forces. The total is Score C.

The third step is Group B: neck, trunk and legs. The neck scores 1 in 0–10° flexion, 2 at 10–20°, 3 above 20° and 4 in extension; +1 each is added for twisting and side bending. The trunk scores 1 when upright or in a well-supported seated posture, 2 in 0–20° flexion, 3 at 20–60° and 4 above 60°; twisting and side bending are again +1 each. The legs score 1 if they are supported and the weight is evenly balanced, otherwise 2. These three values are combined in Table B, and the same muscle use and force adjustments are added to obtain Score D.

In the fourth step, Score C is entered as the row and Score D as the column in Table C, and a final score between 1 and 7 is read off. The action levels corresponding to the final score are shown in the table below.

How is REBA scored step by step?

Step one: in REBA the groups swap places, and Group A is the trunk, neck and legs. The trunk scores 1 when upright, 2 in 0–20° flexion or extension, 3 in 20–60° flexion or extension beyond 20°, and 4 in flexion above 60°; +1 if there is twisting or side bending. The neck scores 1 in 0–20° flexion and 2 in flexion beyond 20° or in extension; twisting or side bending is +1. The legs score 1 if weight is borne on both feet or the person is walking or sitting, and 2 if on one foot or in an unstable posture. If the knees are bent 30–60°, +1 is added, and above 60°, +2; this adjustment is not applied when seated. The three values are combined in Table A and the load score is added: 0 below 5 kg, 1 for 5–10 kg, 2 above 10 kg; +1 if there is a sudden or shock force. The result is Score A.

The second step is Group B: upper arm, lower arm and wrist. The upper arm ranges and their 1–4 scores are the same as in RULA. +1 if the arm is abducted or rotated, +1 if the shoulder is raised; −1 is applied if the person is leaning, the weight of the arm is supported or the arm hangs in the direction of gravity. The lower arm scores 1 in 60–100° flexion and 2 outside this range. The wrist scores 1 in 0–15° flexion or extension and 2 above 15°; deviation or twisting is +1. The values are combined in Table B and the coupling score is added: a well-fitting handle and a mid-range power grip is 0 (good), an acceptable but not ideal hand hold is 1 (fair), a hand hold that is not acceptable but possible is 2 (poor), and no handle or an unsafe one is 3 (unacceptable). The result is Score B.

In the third step, Score A is entered as the row and Score B as the column in Table C. The activity score is added to the value read off: +1 if one or more body parts are held static for longer than 1 minute, +1 if small-range actions are repeated more than 4 times per minute (excluding walking), and +1 if the action causes rapid, large changes in posture or an unstable base. The total is the REBA score, which falls into one of the five risk levels below.

Example: how is lifting a box from the floor scored with REBA?

The following example is hypothetical. A worker picks up 12 kg boxes from a pallet on the floor and places them on a conveyor at waist height; about five boxes are moved per minute and the boxes have no handholes. The moment to be scored is when the worker grips the box on the floor. Scoring proceeds in three steps.

1. Score A (trunk, neck, legs and load): the trunk is bent forward about 70° with no twisting; since this is above 60°, it scores 4. While looking at the box, the neck is beyond 20° relative to the trunk; 2 points. Both feet are on the floor and the knees are bent about 40°; +1 is added to 1, so the legs score 2. In Table A, the intersection of trunk 4, neck 2, legs 2 is 6. Since the load is above 10 kg, +2 is added; Score A is 8.
2. Score B (upper arm, lower arm, wrist and coupling): the arms are about 70° forward relative to the trunk; the upper arm scores 3. The elbows are almost straight, i.e. below 60°; the lower arm scores 2. The wrist is below 15° with no deviation; 1 point. In Table B this intersection is 4. The box has no handholes, but it can be gripped with the fingers by its lower edge; the hand hold is not acceptable but possible, so coupling is poor, +2. Score B is 6.
3. REBA score (Table C and activity): in Table C, the intersection of A 8 and B 6 is 10. Since lifting from the floor to waist height causes a rapid, large change in trunk posture each time, +1 is added for activity. The REBA score is 11: very high risk, action required immediately.

Now suppose we raise the pallet to thigh–waist height with a scissor lift table. The trunk bends 0–20° (2), the neck is below 20° (1) and the knees are straight (1); the Table A value is 2, and with the load Score A becomes 4. Because the box is held close to the body, the upper arm drops into the 20–45° range (2) and the elbow into the 60–100° range (1), while the wrist stays at 1; the Table B value is 1, and with poor coupling Score B is 3. Table C gives 4, and with activity the result is 5: medium risk, action required. A single height adjustment brings the score down two risk levels; the remaining risk is reduced by adding handholes to the box or lowering its weight.

What are the limits of manual assessment?

The first limit is time. The score represents a single moment; how many times and for how long that posture occurs during the shift is reflected only roughly, through the +1 activity or muscle use adjustment. A lift performed 50 times a day and one performed 500 times a day can receive the same score. The second limit is the observer. The developers of RULA report in their reliability study that differences between assessors arose when an angle was on the boundary between two ranges. In the example above, if the upper arm is taken to be hanging in the direction of gravity and −1 is applied, the upper arm drops to 2 and Score B to 4; Table C gives 9, and with activity 10. The same image is very high risk for one assessor and high risk for another. Coupling and force are also open to interpretation. The third limit is sampling: the assessment is usually done on the day shift, with the observer present, over a few cycles; fatigue on the night shift or product changeovers are not captured.

When are OWAS and NIOSH preferred?

OWAS (Ovako Working Posture Analysing System) was developed in 1977 in the Finnish steel industry and also served as a basis for RULA. It codes the back in 4 categories, the arms in 3, the legs in 7 and the load in 3, and assigns each combination to one of 4 action categories. Because it requires no angle measurement, it is suited to making many observations at regular intervals throughout a shift; it gives the distribution of exposure rather than detail.

For lifting tasks, the main reference tool is the NIOSH lifting equation, revised in 1991. The recommended weight limit is calculated with the formula RWL = LC × HM × VM × DM × AM × FM × CM; LC is the 23 kg load constant, and the other multipliers represent the horizontal and vertical position of the load, the lifting distance, trunk twisting, frequency and coupling. The lifting index is the ratio of the load to the RWL; a value above 1 indicates that the task needs to be redesigned. In a task such as lifting a box from the floor, REBA assesses the posture and NIOSH the load; used together, they make clear which adjustment will bring the greater benefit.

How does pose estimation make ergonomic measurement continuous?

In manual assessment a person measures the angle; in pose estimation the camera does. The model extracts the body keypoints in every frame and calculates the trunk, neck and arm angles. In our safe behaviour and pose analysis solution, the body is represented by 17 keypoints and the analysis runs at 25 fps; this means the angle steps of the REBA and RULA worksheets are filled in automatically for every frame. The real gain is duration: how many times a risky posture is repeated during the shift and how long it lasts in total are counted directly, and the effect of an adjustment is measured before and after with the same camera. Inputs that cannot be read from the image, namely load weight, coupling quality and often force, are defined on site per workstation. We describe reporting, anonymity and pilot design in the pose estimation and ergonomics article.

What are the limits of a 2D camera?

A single camera projects the three-dimensional body onto a two-dimensional plane. Movement towards or away from the camera is foreshortened: to a camera looking from the front, a trunk bent forward 60° can look almost like an upright line. For trunk and arm flexion, therefore, the camera should see the person from the side or at a diagonal angle. Overhead cameras largely lose information about bending; the ideal angle for people counting is the weakest angle for ergonomics. Neck and trunk twisting, wrist deviation and forearm rotation are small movements that cannot be measured reliably in 2D images; these adjustments require a second camera or a depth camera. This is why the camera angle is chosen at the survey stage with the ergonomics goal in mind.

Where should you start?

A REBA or RULA assessment on paper is still the right starting point; it quickly shows which workstation is the priority. Two or three workstations with intensive manual handling and a REBA score of 8 or above are good candidates for continuous measurement. At these workstations, comparing an ergonomics specialist's manual assessment with the camera's measurement on the same shift makes visible both the consistency of the method and the duration and repetition that a single moment misses. We explain how ergonomics works on the same infrastructure as other OHS applications in the AI for occupational safety article.

At CX Teknoloji, pose analysis works with existing IP cameras and runs by default on the edge server inside the facility; face recognition is not used, and reports are kept per workstation and shift rather than per person. During the survey we assess your workstations in terms of camera placement and, where suitable, set up a pilot that compares against your manual assessment. For a demo based on a video of one of your workstations, reach us via the contact page.

Frequently asked questions

What is the difference between REBA and RULA?

REBA assesses the whole body, while RULA focuses mainly on the upper limbs. RULA (1993) was developed for fixed-workstation tasks where arm, wrist and neck loading dominate, and gives a score from 1 to 7. REBA (2000) was designed for tasks where the trunk bends, the knees flex, loads are lifted and posture changes; it also factors coupling and activity into the score and gives a score from 1 to 15. If posture varies, REBA is chosen; if it is fixed, RULA.

At what REBA score should action be taken?

Action is required if the REBA score is 4 or above. In the original method, 1 is defined as negligible risk, 2–3 as low risk (action may be required), 4–7 as medium risk (action required), 8–10 as high risk (action soon) and 11–15 as very high risk (action immediately). The score sets the priority; looking at which body part pushed the score up also shows which adjustment will be most effective.

What should be done if the RULA score is 3 or 4?

A RULA score of 3 or 4 puts you at the second action level: further investigation is needed and changes may be required. This score indicates that the posture is outside the suitable range of motion, or that, even if within range, it involves repetition, static load or force. Look at which sub-score is high; if the wrist score is high, tool design is the first candidate for adjustment, and if the neck score is high, screen or part height.

Is a REBA assessment legally mandatory?

It is not mandatory as a specific method; what is mandatory is risk assessment. Law No. 6331 (Türkiye's Occupational Health and Safety Law) obliges the employer to carry out or commission a risk assessment, and the Regulation on Manual Handling Operations also requires the risks arising from manual handling to be reduced. The team carrying out the assessment decides which method is used to measure ergonomic risk; REBA, RULA, OWAS and NIOSH are methods widely used for this purpose.

How is load weight scored in a REBA assessment?

In REBA, the load scores 0 if it is below 5 kg, 1 if it is between 5 and 10 kg, and 2 if it is above 10 kg; +1 is added if the load is applied suddenly or with shock. This score is added to the Table A result for Group A. How the load is held is assessed separately through the coupling score (0–3). In RULA the thresholds differ: 2 kg and 10 kg limits are used, and whether the load is intermittent, static or repeated changes the score.

Can REBA and RULA scores be calculated with a camera?

The angle-based steps can be calculated. A pose estimation model extracts the body keypoints in every frame, measures the trunk, neck and arm angles and converts them into REBA and RULA scores; repetition and duration are counted rather than estimated by observation. Load weight, coupling and often force cannot be read from the image, so they are entered per workstation. The camera should see the person from the side or at a diagonal angle; an overhead camera cannot measure bending.