Migraine assessed by proband report was evaluated in a family study of migraine. A clinical interview of spouses and first-degree relatives by a physician was used as an index of validity. The operational diagnostic criteria of the International Headache Society were used. Of the 378 probands from the general population, 126 had migraine without aura, 127 had migraine with aura, 17 had both migraine without aura and migraine with aura and 108 had never had migraine. Spouses (n = 229) and first-degree relatives (n = 1109) were included in the analyses. Sensitivity, specificity, predictive values and chance-corrected agreement rate for the diagnosis of migraine were 49%, 93%, 81% (PVpos), 77% (PVneg) and 0.47, respectively. The corresponding values for migraine without aura were 58%, 87%, 63% (PVpos), 84% (PVneg) and 0.46 respectively, while the values for migraine with aura were 52%, 88%, 61% (PVpos), 83% (PVneg) and 0.42, respectively. Migraine assessed by proband report is not satisfactory for diagnosing migraine in relatives, since the number of affected relatives is highly underestimated. Our results emphasize the necessity of a clinical interview of the relatives in family studies of migraine.