The relationships between duration of schizophrenic illness before first hospital admission, expressed emotion in key relatives, and illness course have yielded conflicting reports. This study examined the issue from a methodological perspective in a sample of first-episode schizophrenic patients. A 'best estimate' of illness onset, based on a compilation of all sources, was compared with an estimate based on parental report. Parental estimates suggested that the children of high-EE parents had been ill for significantly longer, but this difference was not confirmed by the best estimate. In most cases, the ill children of high-EE parents were living at home before hospital admission and their parents' estimates of duration were quite accurate. In the few cases in which the child was living away from home, high-EE parents strikingly overestimated duration. Children of low-EE families were more likely to be living away from home, but this factor did not explain the consistent underestimates of duration made by their parents. Although duration of illness did not relate to EE, successful engagement in out-patient before in-patient treatment was more common among children from low-EE families.