Physical Therapy Berg Balance Test

What the Berg Balance Test Actually Measures

The Berg Balance Test is a 14-item performance-based assessment that scores functional balance from 0 to 56. Each item is rated on a four-point or five-point ordinal scale depending on the task. It was published by Katherine Berg in 1989 alongside her paper in Physical Therapy journal, and it has since become one of the most widely used tools in geriatric and neurological rehabilitation settings. The test takes roughly 15 to 20 minutes to administer when the patient is cooperative, less if they move slowly. It predicts fall risk in elderly populations with a sensitivity around 0.80 and specificity near 0.75 when using a cutoff score of 45. Scoring works like this: a score below 45 indicates high fall risk, between 45 and 56 suggests low fall risk but some balance impairment, and 56 is a perfect score meaning no balance deficits were observed during the test. The original scoring system gives each item a value from 0 to 4, with a few items maxing out at 3 or 2. You add them all together. That is the entire scoring mechanism. It is not complex, which is partly why it survives in clinical practice despite being decades old.

Physical Therapy Berg Balance Test

Here is how you actually run through the 14 items in the order Berg designed them. The sequence matters because each task builds on postural control demands slightly differently, and reversing the order can change the fatigue profile of the patient. Item one is sitting to standing. The patient starts in a standard chair with arms folded across the chest. They rise without using their hands. Four points means they stand without hand use and do so smoothly. Three points is standing without hands but with unsteadiness. Two points involves one or more attempts with hands still not used. One point means they attempt the stand but cannot do so without hand support. Zero is no attempt possible. Item two is standing unsupported. The patient stands with feet together for two minutes. They hold the position without touching anything. If they need to shift weight or hold onto something, they lose points. Four points is holding for the full two minutes unassisted. Lower scores reflect needing support or being unable to maintain the position for the required duration.

Item three is sitting unsupported. The patient sits with feet flat on the floor and back unsupported for ten minutes. Most people complete this without much trouble. The scoring here is less about endurance and more about maintaining an upright posture without leaning or using arm support. Four points is sitting upright for the full duration. Three points is sitting upright but needing brief rest or slight support. Two points involves periodic postural adjustments. One point is unable to maintain sitting for ten minutes without support. Zero means they require assistance or cannot sit unsupported at all. Item four is standing to sitting. The patient starts from a standing position and lowers themselves into a chair. They are told to sit down smoothly, controlling the descent. This is actually one of the more discriminating items because it loads eccentric control through the lower extremities. A healthy older adult might score four. Someone with quadriceps weakness or knee osteoarthritis often scores two or one because they cannot control the lowering phase. The chair should have a firm seat and straight back. Item five is transferring. This is essentially a sideways transfer from a bed or chair to another surface at the same level. The patient crosses their affected side first if they have a neurological deficit, then pivots and sits down. Four points means they complete the transfer safely with no more than one command needed. Three points is completing it with verbal cues only. Two points requires minimal physical guidance. One point needs moderate assistance. Zero is maximal assistance required or the transfer cannot be completed safely.

Item six is standing with eyes closed. The patient stands feet together, arms at their sides, and closes their eyes for ten seconds. This removes visual input and forces reliance on vestibular and proprioceptive systems. Scoring follows the same pattern: four points is holding steady for the full ten seconds, lower scores reflect swaying, opening eyes early, or needing support. This item catches vestibular deficits that the other standing tasks might miss. Item seven is standing with feet together. The patient stands with feet touching, arms folded, for three minutes. This is a narrow-based stance that is deceptively difficult for many older adults. Four points is maintaining the position for the full three minutes. Three points is holding it with some sway. Two points involves significant sway or brief support needs. One point is unable to hold without support. Zero means they cannot maintain the position at all. Item eight is reaching forward with the outstretched arm while standing. The start position is standing with feet shoulder-width apart. The patient lifts one arm to 90 degrees of flexion, makes a fist with the thumb up, and reaches forward as far as they can without shifting their feet. A ruler is placed at arm's length against the extended hand. Four points means they reach 25 centimeters or more. Three points is 15 to 25 centimeters. Two points is 5 to 15 centimeters. One point is less than 5 centimeters. Zero is unable to reach forward without losing balance.

Item nine is picking up an object from the floor. The patient stands with feet slightly apart and bends to pick up a shoe or similar object placed directly in front of them. They are not allowed to use their hands to support themselves on their legs. Four points means they pick it up easily and steadily. Three points is picking it up with less steady balance. Two points is reaching only partway and regaining balance afterward. One point is unable to pick it up but maintains balance. Zero is unable to attempt the task without falling. Item ten is turning to look behind while standing. The patient stands with feet shoulder-width apart and turns their head and torso to look over each shoulder without stepping or changing foot position. Four points means they looks both sides with appropriate weight shift and maintains good balance. Three points is looking one side well and the other less well. Two points is turning only their head without trunk rotation. One point turns but loses balance. Zero is unable to attempt the task. Item eleven is standing with one foot elevated. The patient lifts one foot and places it on a step or box approximately 15 centimeters high, holds for 30 seconds, then switches. This tests single-leg stability. Four points is balancing with appropriate arm use and steady posture. Three points is minor balance adjustments. Two points is moderate adjustments but completing the time. One point is unable to complete the full 30 seconds. Zero is unable to attempt the task.

Item twelve is standing on one leg. The patient lifts one leg and stands on the other for 30 seconds. This is the tandem stance equivalent but on a single leg. Four points is holding for the full 30 seconds. Three points is 15 to 29 seconds. Two points is 5 to 14 seconds. One point is less than 5 seconds. Zero is unable to lift the foot or attempt the task. Item thirteen is sitting with eyes closed. The patient sits in a chair with feet flat and closes their eyes for ten seconds. This isolates postural control in sitting without visual input. Four points is maintaining sitting balance for the full duration. Lower scores reflect swaying, opening eyes early, or falling out of the chair. This item is often skipped in rushed administrations because it adds only a few minutes, but it provides useful information for patients with known vestibular disorders. Item fourteen is standing from a chair without using arms. The patient sits in a chair and stands up using only their legs, keeping their arms crossed or at their sides. This is similar to item one but performed from a different starting position and often used as a retest or alternative depending on the protocol. Four points means they stand smoothly without arm use. Lower scores reflect reliance on arm support or inability to complete the task.

How It Works in Real Clinical Practice

I spent about six years running this test daily in an outpatient clinic, mostly with stroke patients and older adults with balance complaints. The test itself is straightforward on paper, but the real world introduces complications that the manual does not address. One thing beginners consistently get wrong is how they phrase instructions. Berg wrote very specific scripted instructions for each item, and deviation from those scripts can affect the reliability of the scores. If you tell a patient to "stand up carefully" instead of following the exact script, you are introducing variability that is not part of the normed data. Another issue that nobody warns you about is the chair. The standard Berg protocol assumes a chair with armrests removed and a firm, non-rolling seat. I once administered the test in a clinic that used standard office chairs with wheels and padded seats. Three patients used the armrests during the sitting-to-standing task and scored higher than they actually could. Another three couldn't get enough leverage on the soft seat and scored lower. We ended up buying a set of basic dining chairs from a thrift store and keeping them specifically for balance testing. It cost us about forty dollars per chair and eliminated a whole class of measurement error. A more specific edge case I ran into involves patients with unilateral knee replacements. Item seven, standing with feet together, and item twelve, standing on one leg, can produce artificially low scores not because of a true balance deficit but because the patient has pain or mechanical limitation in the knee. I had a patient who scored 41 on the Berg, which would normally indicate high fall risk. When I dug into the item breakdown, her scores were dragged down entirely by items seven, twelve, and nine, all of which loaded heavily on her right knee. She had no other balance complaints and managed her community independently. I documented the confounding factor and adjusted the clinical interpretation, noting that her overall balance was likely in the low-risk range once the orthopedic limitation was accounted for. This is a documented problem in the literature but rarely addressed in the training materials. If a patient's Berg score seems inconsistent with their functional status, you need to check whether joint pathology or pain is driving the score rather than true postural instability.

What the Berg Test Misses

The biggest limitation of the Berg Balance Test is that it only goes up to 56 points, which creates a ceiling effect. Patients who are highly functional, such as those recovering from minor strokes or younger individuals with neurologic conditions, often score 54 or 55 and there is no way to differentiate their balance capacity further. I encountered this frequently with post-polio patients and young adults with mild cerebellar ataxia. Their Berg scores clustered at the top, giving a false impression that the test had no discriminative power for that population. For these patients, the Timed Up and Go test or the Balance Evaluation Systems Test provides better granularity in the high-functioning range. The test also does not assess dynamic balance during walking, which is where most community falls actually occur. A patient can score 50 on the Berg and still trip over a rug at home because the test never challenges gait while performing a dual task. The Four Square Step Test or the Dynamic Gait Index fills that gap. I stopped relying solely on the Berg about five years ago once I realized that my fall prediction model had a blind spot. Combining the Berg with a timed gait measure improved my fall risk stratification significantly. Another issue is the two-minute duration of the standing unsupported item. For patients with cardiovascular compromise or severe deconditioning, holding a standing position for two minutes is physiologically stressful and can produce inaccurate results due to fatigue rather than true balance impairment. In those cases, I shorten the timing and note it in the chart, or I use the Sit-to-Stand timed test instead as a substitute measure. This is not in the original manual but it is a practical adjustment that experienced clinicians make routinely.

Administration Details and Scoring Conventions

You need a standard chair without armrests, a stopwatch, a ruler marked in centimeters, a shoe or small object for the floor pickup item, a step or box approximately 15 centimeters high, and a tape measure. The environment should be a clear space free of clutter with enough room for the patient to move safely. Having a second clinician present during the administration is standard practice, primarily for safety. If the patient begins to sway excessively, you place your hands near their trunk as a protective standby without touching them unless necessary. Actual contact during the test disqualifies that item score unless the patient actually falls or uses support, in which case you score based on what they did, not what you prevented. One scoring nuance that gets overlooked: for item five, the transfer, the direction of transfer matters for neurological patients. Berg's original protocol specifies that patients with hemiparesis should transfer toward their unaffected side first. If you let them transfer toward their weak side, the score will be lower even though their balance capacity is unchanged. This is a common administration error that I caught repeatedly when I started training new therapists. The manual mentions it but the detail is easy to miss if you are reading quickly. For the reaching item, make sure the patient does not lean their entire trunk forward instead of extending their arm. The instruction is to reach with the outstretched arm while keeping the feet planted. Trunk lean counts as a shift in center of gravity and should not be permitted. I mark this clearly in my protocol sheet and give patients a brief practice trial before scoring begins. The practice trial does not count toward the score.

Where to Get the Test Materials

The Berg Balance Test itself is not copyrighted in a way that prevents clinical use. You can download free scored forms and instruction sheets from several university rehabilitation department websites. The original scoring sheet is available through the University of Alberta's occupational therapy resources and through various physical therapy association portals. I use a modified version that includes a section for noting confounding factors like joint pain, fear of falling, or assistive device use during the test. This addition does not change the scoring but improves the clinical utility of the record. The base test materials, including the official scoring form and administration guidelines, are freely downloadable from academic health science center websites without requiring payment or license agreements. Some commercial publishers offer pre-bound Berg test kits with plastic score cards, marked rulers, and laminated instruction sheets. These run anywhere from fifteen to forty-five dollars depending on what is included. The printed materials are fine but not necessary. A printed copy of the instructions, a standard ruler, and a spreadsheet for recording scores is sufficient for most clinic settings. I have found that having a single master copy of the test in a clinic binder with a pen clipped to it lasts for years and costs virtually nothing to maintain.

Interpreting Scores in Context

A Berg score of 56 does not mean a patient is immune to falls. It means they performed well on fourteen specific tasks in a controlled clinic environment. A score of 45 is a common cutoff used in research, but in practice, I look at the pattern of scores across items rather than the total alone. A patient who scores 50 but loses three points on items involving single-leg stance and forward reaching has a different risk profile than a patient who scores 50 but loses points only on the eyes-closed items. The first patient likely has a proprioceptive or strength deficit. The second is more likely to have a sensory integration issue related to vision dependency. The test has been validated across multiple populations including stroke, Parkinson's disease, multiple sclerosis, and community-dwelling elderly. The predictive validity for falls in hospital inpatients is weaker than in community-dwelling older adults, with area under the curve values ranging from 0.65 to 0.80 depending on the population studied. For stroke patients specifically, a cutoff of 45 still holds reasonably well, but the test tends to underestimate balance ability in patients with mild impairments who have compensated well through assistive devices or environmental modifications. If you are using the Berg Balance Test as part of a discharge planning process, combine it with at least one functional mobility measure and a history of recent falls. The combination yields better predictive accuracy than the Berg alone. I have seen too many patients discharged home based on a borderline Berg score without checking whether they could navigate stairs or perform a dual-task walk. Those are the patients who end up back in the hospital within thirty days.

The test remains useful because it is simple, reliable, and correlates well with real-world function when interpreted correctly. It is not a standalone diagnostic tool and it should never be treated as one. Most clinics that use it properly administer it once at intake, once at discharge, and sometimes at intermediate intervals for patients with progressive or fluctuating conditions. The retest interval depends on the diagnosis and the rate of expected change, not on a fixed schedule. For stroke rehabilitation, a three-week interval between assessments is typical. For fall risk screening in assisted living, an annual reassessment is adequate for stable residents.