Amitriptyline is one of the most commonly used medications in migraine prophylaxis. There have been relatively few placebo-controlled studies of amitriptyline in migraine prophylaxis or in treatment of chronic daily headache (CDH). This report deals with a large placebo-controlled trial of amitriptyline vs placebo of 20 weeks duration that included subjects with intermittent migraine (IM) as well as CDH. The study was carried out between 1976 and 1979; however, results have never been fully reported. Patients with a history of migraine as defined by the 1962 Ad Hoc Committee report were recruited for this study. Subjects had at least 2 headaches per month, and no limit was placed on the number of headaches per month that could be experienced. The study format included a 4-week baseline period (Period A) in which all subjects received placebo in a dose of 2 pills per day for one week, 3 pills per day for one week and then 4 pills per day for 2 weeks. Subjects with at least 2 migraine headaches in this period were then entered into Period B and randomized into either amitriptyline or placebo tracks. Medication consisted of identical tablets containing either 25 mg amitriptyline or placebo. Period B was 4 weeks in duration with dose titration identical to Period A. The dose could be reduced if necessary to reduce side effects. The minimum dose was one pill per day. Period C was a 12-week maintenance or stabilization period in which the patient continued the dose established by week 8 with visits at weeks 12, 16, and 20. Patients kept a headache calendar that was used for data collection. Headache frequency (per month), severity, and duration (hours) were the primary measurement parameters employed for data analysis. For the entire group, 391 subjects were entered into Period A, 338 were randomized into Period B, 317 (81%) subjects completed the first post-randomization visit (8 weeks), 255 (65%) completed week 12, 210 (54%) completed week 16, and 186 (48%) completed week 20. Using headache frequency and evaluating parameters of (a) improvement, (b) no change, or (c) worsening relative to baseline, there was a significant improvement in headache frequency for amitriptyline over placebo at 8 weeks (P = .018) but not at 12, 16, or 20 weeks. When amitriptyline and placebo patients were compared for headache frequency at 8, 12, 16, and 20 weeks to their own placebo stabilization period at 4 weeks, statistically significant improvement vs worsening was seen in headache frequency at each evaluation point for both amitriptyline and placebo groups (P ≤ .01) reaching 50% reporting a decrease in frequency in each group and approximately 10% reporting worsening by week 20. There were no significant differences in headache severity or duration between amitriptyline and placebo groups at anytime during the study. Within the study sample, there were 36 amitriptyline and 22 placebo subjects who had headaches ≥ 17 days/month that fit the current definition of CDH by the Silberstein-Lipton criteria. These were analyzed separately as a subgroup for comparison of amitriptyline vs placebo using a metric of (1) no change or worsening; (2) up to a 50% improvement; and (3) ≥ 50% improvement in headache frequency. Amitriptyline was superior to placebo in number with improvement in frequency of ≥ 50% at 8 weeks (25% vs 5% [P = .031]) and at 16 weeks (46% vs 9% [P = .043]). There was a trend for amitriptyline to be superior to placebo at 12 and 20 weeks but this did not reach significance. In this study, using headache frequency as the primary metric, for the entire group, amitriptyline was superior to placebo in migraine prophylaxis at 8 weeks but, because of a robust placebo response, not at subsequent time points. For the subgroup with CDH, amitriptyline was statistically significantly superior to placebo at 8 weeks and 16 weeks with a similar but nonsignificant trend at 12 and 20 weeks. Compared with placebo amitriptyline is effective in CDH. Amitriptyline was also significantly effective in IM compared intragroup to its own baseline; however, placebo was equally effective in the same analysis. The reason for the robust placebo response in the IM group is not clear, but has been occasionally reported.