The definition of efficacy parameters in drug trials of migraine has received little attention and the tendency has been to take into account many parameters, which leads to the possibility of some positive results being obtained purely by chance. In trials of prophylactic drug treatment, the major parameter is the frequency of attacks. Patients with frequent interval headache should preferably be excluded from such trials since there is no satisfactory way to distinguish interval headache from mild migraine attacks. Severity and duration of attacks are important parameters but they are difficult to interpret since they are modified by symptomatic medication. The mean score of each attack by consumption of symptomatic medication gives a good estimate of the severity and duration of attacks, but the problem of equalizing different medications in order to obtain a single estimate has not been solved. In trials of abortive drug treatment, there is at present no agreement upon the single most important parameter. Severity of headache, severity of associated symptoms, global efficacy and duration of attacks are all modified by escape medication. The need for escape medication is a better parameter but its power to discriminate ineffective and partially effective drugs is low. It is suggested that the major parameter could be the number of attacks aborted in 2 h or less, which is a simple quantitative and clinically relevant criterion.