Objective To assess validity of ICHD-3 diagnostic criteria for menstrual migraine. Methods We performed a longitudinal E-diary study in premenopausal women with migraine. Menstrual migraine diagnosis was self-reported at baseline, and verified according to diary based ICHD-3 criteria and a previous proposed statistical model. Validity of self-reported menstrual migraine was compared to diary based diagnosis and statistical diagnosis. Test-retest reliability and concordance between both methods were determined. Clinical characteristics of perimenstrual and non-perimenstrual migraine attacks were compared in women with and without menstrual migraine. Results We included 607 women. Both women who did and women who did not self-report to suffer from menstrual migraine fulfilled ICHD-3 criteria in the E-diary in two thirds of cases. Pure menstrual migraine was extremely rare (<1%). Concordance between statistical and diary based diagnosis was minimal (κ = 0.28, 95% CI:0.23–0.33). Women diagnosed with menstrual migraine showed 37–50% longer attack duration and increased triptan intake (OR 1.19–1.22, p < 0.001) during perimenstrual attacks. Conclusion Self-reported menstrual migraine diagnosis has extremely poor accuracy. Two thirds of women suffer from menstrual migraine, independent of self-reports. Pure menstrual migraine is rare. Women with menstrual migraine have longer attack duration and increased triptan intake during perimenstrual attacks, in contrast to women without menstrual migraine. Prospective headache (E-)diaries are required for a menstrual migraine diagnosis, also in clinical practice.