The Modified Ashworth Scale (MAS) is a clinical scale used to assess muscle spasticity. While the evidence indicates that the reliability of the MAS is better in the upper limb and in certain distal muscle groups, no investigation has compared the effect of limbs and muscle groups on the MAS reliability. This study aimed to evaluate the effect of limb and muscle group on the reliability of the MAS in patients with spastic hemiplegia. Thirty subjects with upper and lower limb muscle spasticity were recruited for this trial. Two female experienced physiotherapists participated in this examination of reliability, and rated each patient in a randomized order in a single session. For the intrarater reliability, the second rater repeated the test 1 week later. Shoulder adductor, elbow flexor, wrist flexor, hip adductor, knee extensor, and ankle plantar flexor were tested on the hemiplegic side. Results demonstrated moderate inter (κ = 0.514, SE = 0.046, p < 0.001) and intrarater (κ = 0.590, SE = 0.051, p < 0.001) reliability. For the inter and intrarater reliability, the agreement obtained for the upper and lower limb was similar. In the upper limb, the agreement between raters on the distal wrist flexor was significantly higher than the agreement on the proximal shoulder adductor. In the lower limb, there was a similar agreement between raters on the distal ankle plantar flexor and proximal hip adductor. For within rater, the agreement on the proximal and distal muscles of both limbs was not statistically significant. The Modified Ashworth Scale had moderate reliability. The limbs had no effect on the reliability. The agreement on distal wrist flexor in the upper limb was significantly higher between rater than in the proximal shoulder adductor. The agreement obtained with the MAS was not good, which questions the validity of the measurements.