This study examined evidence for the role of selective memory for stressful events in women with headache. Previous studies have reported that on retrospective measures of stress, those with tension-type headache report more stressful events and rate the events as more stressful than did headache-free controls. However, when ratings are made concurrently, participants with headache and controls did not differ in their ratings of equivalent stressors presented in the laboratory. One theory for why differences are found in stress ratings made retrospectively, but not concurrently, is that selective memory biases recollection of past events in patients with headache. This study compared self-report ratings of stressful events and their perceived impact made either concurrently or retrospectively to determine if selective memory might explain the discrepancies found in earlier studies. Participants included 20 patients with tension-type headache and 22 headache-free controls. Participants were compared on hourly, daily, and weekly measures of stressors and their perceived impact via hourly and nightly visual analog scale ratings, the Daily Stress Inventory, and the Weekly Stress Inventory. If support was to be offered to the hypothesis that selective memory biases the retrospective memory of patients with headache, then an interaction between group assignment (ie, headache versus control group) and time of rating would be expected. No such interaction occurred. Results from the study suggest that all participants have a tendency to overestimate stress on retrospective measures, but that patients with headache do not do so at a significantly different rate than do headache-free controls. The alternative hypothesis that patients with headache tend to appraise everyday events as more stressful is supported.