Modified Ashworth Scale: Grades, Reliability, and Limits

The Modified Ashworth Scale is a clinical grading system used to rate muscle spasticity on a six-point scale from 0 (no increase in tone) to 4 (the limb is rigid). It is the most widely used bedside tool for measuring how much resistance a muscle puts up when a clinician moves it passively through its range of motion. Despite decades of use in stroke rehabilitation, cerebral palsy management, and spinal cord injury care, the scale has well-documented reliability problems and a structural quirk that still causes confusion among clinicians.

What the Grades Mean

The original Ashworth Scale was a five-point tool published in 1964. In 1987, Bohannon and Smith added a “1+” grade between the original 1 and 2 to capture patients who had a slight increase in tone with a catch followed by minimal resistance through the rest of the range. That single addition created the Modified Ashworth Scale, and it was this version that showed good interrater reliability when two clinicians independently graded elbow flexor spasticity in 30 patients with brain injuries, agreeing on roughly 87% of their ratings.¹ The grades run as follows:

0: No increase in muscle tone.
1: Slight increase in tone, with a catch and release or minimal resistance at the end of the range of motion.
1+: Slight increase in tone, with a catch followed by minimal resistance through less than half of the range.
2: More marked increase in tone through most of the range, but the limb is still easily moved.
3: Considerable increase in tone; passive movement is difficult.
4: The affected part is rigid in flexion or extension.

The examiner moves the limb through its full available range while the patient tries to stay relaxed. The entire motion is supposed to take about one second, though the original description did not specify speed precisely, a gap that has caused persistent trouble.

The 1+ Problem

The grade that was meant to improve the scale’s sensitivity is also the grade that causes the most headaches. The distinction between a score of 1 and a score of 1+ hinges on whether resistance persists through “less than half” of the range after an initial catch. In practice, deciding exactly where the halfway point is, especially on a limb that may not have full range to begin with, introduces subjectivity that different clinicians resolve differently. One analysis of four stroke survivors all graded at 1+ found a highly varied set of signal outputs when the same passive stretch was measured with sensors, meaning the 1+ bin was grouping together people whose muscles were behaving quite differently.²

A widely cited review concluded that until the ambiguity between the 1 and 1+ grades is resolved, the MAS should be treated as measuring resistance to passive movement at a nominal level rather than on a true ordinal scale.³ In plain terms, that means you cannot assume the jump from 1 to 1+ represents the same size of change as the jump from 2 to 3. Researchers who run statistics on MAS scores as if the numbers are evenly spaced may be drawing conclusions the scale’s structure does not actually support.

How Reliable Are the Ratings

Reliability studies on the MAS paint a picture that depends heavily on which muscle you are testing and who is doing the testing. The original 1987 study found strong agreement at the elbow flexors, and that finding has held up fairly well for that specific muscle group. But the elbow is one of the easier joints to test; the patient’s arm rests naturally, and the examiner has good mechanical control of the movement.

When researchers looked at a broader set of muscles, the numbers dropped. A study testing multiple upper- and lower-limb muscle groups found moderate interrater reliability overall, with better agreement at the distal wrist flexors than at the proximal shoulder in the upper limb.⁴ In children with increased muscle tone, interrater reliability was good for elbow flexors and hamstrings but poor for other muscles, and even within the same rater, consistency was only moderate for most muscle groups outside the hamstrings.⁵

In people with spinal cord injuries, the picture is especially messy. One study found that reliability depended enormously on the statistical method used: weighted kappa values looked acceptable, but simple kappa values fell into the unreliable range for several muscle groups. The authors recommended that repeated assessments always be done by the same rater and in a highly standardized way.⁶ When a pilot study tried to improve consistency by standardizing the speed of passive movement, the results were still only moderate for the wrist and slight for the ankle.⁷

The upshot is that two different clinicians testing the same patient on the same day can easily disagree by a full grade, especially at joints other than the elbow. That matters because many treatment decisions, including whether to adjust a medication dose or authorize further botulinum toxin injections, rest on whether MAS scores have changed by one or two points.

Does It Actually Measure Spasticity

Spasticity, as it was defined in the landmark 1980 description by neurophysiologist James Lance, is a velocity-dependent increase in stretch reflexes. The key word is velocity-dependent: a truly spastic muscle resists more when you move it faster. The MAS, however, asks the clinician to move the limb at roughly one consistent speed and rate the resistance they feel. That means the scale captures resistance to passive movement but does not specifically isolate the velocity-dependent neural component from other sources of stiffness.

This distinction matters because increased resistance can also come from contracture, where the muscle or connective tissue has physically shortened or stiffened over time, from fibrosis, or from changes in the mechanical properties of the muscle itself. A study comparing the MAS with the Tardieu Scale found that in every case where the MAS overestimated spasticity, the patient had a contracture. The MAS could not distinguish the neural component from the structural one.⁸

Electromyography studies tell a mixed story about how well the scale tracks what muscles are actually doing electrically. In patients with tetraplegia, each increasing MAS grade corresponded to increasing electrical activity in the muscle, and the duration of that activity correlated consistently with MAS grades.⁹ A separate study of hemiparetic stroke patients also found a positive correlation between the MAS and both the magnitude and duration of the muscle’s electrical response to stretch.¹⁰ These findings suggest the scale is picking up something real about neural-driven muscle activation.

But when researchers used more sophisticated biomechanical setups to separate reflex-driven stiffness from intrinsic mechanical stiffness, the correlation with MAS grades broke down. One study found no significant correlation between the MAS and either muscular torque, reflex torque, or the stiffness components measured by their device.¹¹ Meanwhile, a different biomechanical model did find that total resisting force correlated with MAS scores.¹² And when resistance torque was measured mechanically in patients with traumatic brain injury, the correlation with MAS scores was fair but not strong.¹³

Reading all of this together, the MAS appears to capture overall resistance to passive movement reasonably well, and that resistance does partially reflect spasticity. But the scale cannot separate spasticity from contracture, fibrosis, or other mechanical changes in the tissue. For clinicians making treatment decisions that depend on the cause of stiffness, not just the amount, the MAS alone is not enough.

How It Is Used in Treatment

Despite its limitations, the MAS remains the standard outcome measure in clinical trials of anti-spasticity treatments. Botulinum toxin studies routinely use it as the primary or co-primary endpoint. An umbrella review of meta-analyses found that botulinum toxin injections produced a significant reduction in MAS scores across the vast majority of studies, with the effect consistently favoring the treatment.¹⁴ The road to regulatory approval was not entirely smooth: the FDA initially refused to accept the Ashworth Scale as a primary endpoint for botulinum toxin trials, concerned about the scale’s measurement properties. Several randomized, placebo-controlled trials eventually demonstrated significant improvements in MAS scores alongside patient treatment goals, which together persuaded the FDA to approve onabotulinumtoxinA for adult upper limb spasticity.¹⁵

In observational practice, the MAS is used to track response to botulinum toxin in patients with spasticity from multiple sclerosis, spinal cord injury, and traumatic brain injury alike.¹⁶

Intrathecal baclofen therapy, where a surgically implanted pump delivers the drug directly into the spinal fluid, also uses the MAS to judge outcomes. In one series of 112 patients with severe spasticity, mean MAS scores dropped from about 4.5 before the pump was placed to about 1.2 on chronic intrathecal baclofen.¹⁷ A systematic review and meta-analysis of intrathecal baclofen pump therapy confirmed a significant decrease in MAS scores in both upper and lower limbs.¹⁸ In a smaller series of 16 patients, average MAS scores fell from about 3.7 before surgery to about 1.8 two weeks after, with the improvement holding at long-term follow-up.¹⁹

Challenges in Children with Cerebral Palsy

Cerebral palsy is one of the most common clinical settings for MAS use, and it is also one of the settings where the scale performs most inconsistently. A study comparing the Ashworth and Modified Ashworth Scales in children with spastic cerebral palsy concluded bluntly that these scales are not very reliable and that assessments should be interpreted with great caution.²⁰ A separate study testing four lower-limb muscle groups in children with spastic cerebral palsy found that interrater reliability for both the MAS and the Modified Tardieu Scale fell below the commonly accepted threshold for clinical use.²¹

Why does the scale struggle with children? Several factors converge. Young children may not cooperate with the request to relax. Their smaller limb size makes it harder for the examiner to control movement speed and isolate resistance. And the distinction between spasticity and dystonia, another common motor pattern in cerebral palsy involving involuntary sustained muscle contractions, is particularly difficult to parse through passive stretch alone. A preliminary study of the “Modified Modified Ashworth Scale” (MMAS), which removes the troublesome 1+ grade and collapses the scoring into a cleaner ordinal scale, showed excellent reliability for lower extremity spasticity in children with cerebral palsy, though the authors cautioned that small sample sizes and wide confidence intervals limit how firmly you can rely on that result.²²

The Tardieu Scale as an Alternative

The main competitor to the MAS in clinical spasticity assessment is the Tardieu Scale, or its updated version, the Modified Tardieu Scale. The Tardieu approach tests resistance at two different speeds: a slow passive stretch and a fast stretch. By comparing the response at each speed, the clinician can separate velocity-dependent resistance (spasticity) from velocity-independent resistance (contracture or tissue stiffness). That structure aligns more closely with the definition of spasticity as a velocity-dependent phenomenon.²³

In patients with severe brain injury and impaired consciousness, the Modified Tardieu Scale has demonstrated higher test-retest and interrater reliability than the MAS.²⁴ Some studies have also found the Tardieu Scale to be more sensitive to changes following botulinum toxin treatment. Yet despite these theoretical and empirical advantages, the Tardieu Scale has not displaced the MAS in widespread clinical use. It takes more time to administer, requires the clinician to perform two separate stretches at controlled speeds, and does not have the same depth of accumulated trial data. Most botulinum toxin and intrathecal baclofen trials were designed around the MAS, so switching primary endpoints mid-field would create problems of comparability.

Where MAS Scores Miss What Patients Care About

One of the less discussed limitations of the MAS is that a change in score does not automatically translate into a change in function. A patient whose elbow flexor spasticity drops from a 3 to a 1 after botulinum toxin may have a limb that is mechanically easier to move, but that does not guarantee they can now dress themselves, reach for objects, or walk more comfortably. Spasticity is only one contributor to disability in conditions like stroke, and muscle weakness, loss of coordination, and sensory deficits all play roles that the MAS does not capture.

A large study of people more than a year after ischemic stroke found that about a third had increased muscle tone, with upper limbs affected more often than lower limbs.²⁵ But tone is just one piece of the functional puzzle. Research on patients with HTLV-1-associated myelopathy found that muscle weakness, not spasticity, was the dominant factor limiting walking ability, functional independence, and quality of life.²⁶ Clinicians increasingly pair MAS assessments with patient-reported outcome measures and goal-attainment scaling to capture whether reduced tone actually improves daily life.

Technology-Driven Alternatives on the Horizon

The limitations of subjective bedside grading have pushed researchers toward instrumented approaches. Portable devices, robotic equipment, and myotonometry tools can now use angular, inertial, torque, and surface electromyography sensors to quantify spasticity, often with the help of machine learning algorithms to interpret the signals.²⁷ Some groups have built quantitative models that map sensor data onto MAS grade descriptions, aiming to replicate what a clinician’s hands feel but with continuous numerical output instead of a six-point ordinal scale.²⁸

These tools solve several problems at once. They remove the subjective element of grading. They generate continuous data instead of coarse ordinal categories, letting researchers detect smaller changes. And many of them can separate the neural reflex component from passive tissue stiffness, the exact distinction the MAS fails to make. The trade-offs are cost, setup time, and the fact that most of these devices exist only in research settings. None has achieved the ubiquity of a clinician’s two hands and a memorized grading table, and for a bedside screening tool, simplicity is a genuine advantage. It seems likely that the MAS will remain the workhorse of clinical practice for some time, with instrumented tools supplementing rather than replacing it for patients and research questions that demand higher precision.