We tested the hypothesis that changes in soleus stretch reflex was correlated to changes in intrathecal baclofen dose in 12 multiple sclerosis patients with moderate-severe spasticity treated with intrathecal baclofen pump. Twice patients were evaluated clinically and biomechanically. The short-latency soleus stretch reflex was elicited by rotating the ankle joint 4 degrees with a velocity from 3.1 to 180 degrees/s. There was a strong correlation between changes in intrathecal baclofen dose and amplitude of the short-latency stretch reflex (r=-0.88, P<0.001), which means that with an increase in baclofen dose there is a decrease in the amplitude. In contrast, no correlation exists between changes in intrathecal baclofen dose and clinical assessment of spasticity by using the Ashworth scale. The amplitude of the stretch reflex was very small (5 microV) compared with previous findings (>50 microV), which indicates an effective antispastic effect of intrathecal baclofen. We suggest that clinical evaluation of spasticity using Ashworth scale is insensitive to detect minor changes in moderate-severe spasticity and consequently might not be very useful in evaluating spasticity in relation to ambulatory filling of baclofen pumps. The soleus stretch reflex might be useful in situations when there is doubt about the effect of intrathecally administered baclofen.