In 20 patients with pure menstrual migraine either Estraderm TTS 50 patches (E) or placebo (P) patches were applied during three successive menstrual cycles, randomly allocated to the treatment sequences E-P-E or P-E-P. Clinical neurophysiological tests, contingent negative variation (CNV) and exteroceptive temporalis muscle suppression test (ETST) were performed before treatment. The predictive value of these tests regarding the efficacy of Estraderm TTS was studied. Neither the number, duration and severity of the migraine attacks, nor the consumption of analgesics and ergotamine differed significantly during Estraderm TTS and placebo treatment. The ETST was consistent with migraine in 35% (95% confidence interval 15.4 to 59.2) of the patients. The CNV in 55% (31.5 to 76.9), and both tests in 25% (8.7 to 49.1%). Regarding the prediction of the Estraderm TTS effect on the migraine attacks, the specificity of the ETST, CNV and the combination of ETST with CNV, calculated for the first two cycles, was respectively 81.8% (48.2 to 97.7), 45.5% (16.8 to 76.6) and 80% (28.4 to 99.5). For the last two cycles these values were respectively 75% (42.8 to 94.5), 50.0% (21.1 to 87.9) and 71.4% (29.0 to 96.3). The sensitivity of the tests was respectively 62.5% (24.5 to 91.5), 62.5% (24.5 to 91.5) and 66.7% (22.3 to 95.7) in the first 2 cycles. In the last 2 cycles 50.0% (15.1 to 84.3), 62.5% (24.5 to 91.5) and 60% (14.7 to 94.7). This study did not demonstrate an effectiveness of Estraderm TTS in perimenstrual migraine, except for the subgroup of perimenstrual migraine patients in whom the ETST test results were consistent with migraine.