Accurate prognosis of motor recovery assists rehabilitation planning and efficient resource allocation by clinicians and patients. However, the relationship between initial impairment and subsequent recovery is highly variable, making accurate prognosis for individual patients difficult. Neurophysiological and neuroimaging techniques offer more predictive power than clinical assessment, and can be used to identify patients with previously unrecognised potential for recovery. We have recently developed the PREP algorithm for Predicting Recovery Potential after stroke, which uses TMS to test the functional integrity of descending motor pathways to the paretic upper limb. The algorithm had positive predictive power of 88%, negative predictive power of 83%, specificity of 88% and sensitivity of 73%. PREP is currently being evaluated in a 3-year pragmatic clinical trial of patients admitted to Auckland City hospital with stroke and upper limb weakness. Data acquired in the first 12 months confirm an essential role for TMS in the prognosis of recovery of upper limb function. We use clinical, neurophysiological and neuroimaging measures to predict the recovery of upper limb function. Upper limb impairment is evaluated within 72 h of stroke by grading shoulder abduction and finger extension strength. Functional integrity of descending motor pathways is evaluated within 10 days after stroke by determining whether motor evoked potentials could be elicited from the paretic wrist extensor muscles. Structural integrity of the posterior limbs of the internal capsules is evaluated with diffusion-weighted magnetic resonance imaging, conducted within 14 days after stroke. These measures are combined in the PREP algorithm to predict paretic upper limb function at 12 weeks after stroke for each patient. To date, we are finding further evidence that a combination of clinical and TMS assessment can provide an accurate prognosis for 80% of the patients studied, with MRI required to resolve uncertainty for the remaining 20%. This highlights the valuable information provided by a relatively simple neurophysiological assessment. TMS is a useful tool for predicting recovery of upper limb function after stroke, particularly in combination with clinical and imaging measures, as part of the PREP algorithm. The clinical benefits of tailoring rehabilitation based on the individual’s capacity for motor recovery will be discussed.