While headache represents one of the most common medical complaints of children, very little is known about its phenomenology. In the two studies currently reported, children's headache complaints were assessed in several ways. In Study 1, 53 children and their parents provided estimates of 3 parameters of headache during separate interviews and then engaged in 4 continuous weeks of daily record keeping, all of which occurred prior to participating in a behavioral treatment program. In Study 2, 37 of the children who had completed treatment and their parents participated in the same type of assessment to determine correspondence between various ways to gauge improvement. Additional estimates of improvement were collected by administering visual analog scales to children, parents, and a pediatric neurologist (upon completion of his independent interview). Although child and parent estimates of headache activity obtained at the initial interview were highly consistent, these estimates greatly overstated the values recorded by the children in their daily diaries. Interview estimates overstated diary values by from 56% to 112%. The interview estimation biases were greatly reduced by treatment's end, however, and were no longer significant for all but 1 measure. Parent and child diaries revealed high levels of agreement for headache intensity both prior to and following treatment. Signficant disagreement occurred prior to treatment for headache frequency, but this did not continue following completion of treatment. Visual analog estimates of improvement were not statistically different from improvement measured by diary report, although children, parents, and neurologist underestimated actual improvement by 5%, 10%, and 20%, respectively.