OBJECTIVE To correlate repetitive nerve stimulation (RNS) decrement in different muscles with the predominant clinical presentation in myasthenia gravis (MG), and to study single fibre EMG (SFEMG) sensitivity in ocular MG. METHODS Sixty-nine, untreated, consecutive patients suspected for MG were observed prospectively for a minimum of 6 months. Those who improved on medical treatment were diagnosed as MG. The others, in whom the neurophysiological studies were normal and that did not improve on medical treatment served as a control group, from which normative data for RNS and SFEMG was obtained. The MG patients were further classified in 3 subgroups according to the predominant clinical presentation: group I (ocular); group b (bulbar); and group a (axial). We performed RNS in nasalis, trapezius, anconeus, and abductor digiti minimi. All patients with ocular MG underwent jitter determination of the orbicularis oculi muscle. RESULTS Thirty-seven patients were diagnosed as MG (group I, 15; group b, 13; group a, 9). In group I, RNS was abnormal in 33% of the patients. RNS studies disclosed at least one abnormal muscle response in every patient in groups a and b. Trapezius was significantly more sensitive in group a, and anconeus and nasalis in group b (P < 0.01). Jitter was abnormal in all patients in group I, and the most sensitive parameter was an increased number of unstable pairs, 100%. CONCLUSIONS Based on these observations, we recommend that a shoulder muscle, as the trapezius, should be studied first in the limb-axial presentation of MG, and the anconeus-nasalis muscles in predominant bulbar MG. In ocular MG, RNS is not sensitive and jitter should be performed in facial muscles. SIGNIFICANCE This paper shows the unequal sensitivity of several muscles to RNS in different forms of MG.