Epidemiological studies often use age-of-first-admission from psychiatric case registers to estimate age-of-onset in schizophrenia. Retrospective, interview-based methods have shown that there is a delay between onset of symptoms and eventual contact with psychiatric services, and that this delay can vary both among individuals and at different ages. This delay or lag can confound the interpretation of first admission data such as age-of-onset. To evaluate the potential impact of this factor, we constructed a flexible mathematical model which integrates age-at-first-admission with estimates of this lag, which were derived from interview-based studies and clinical judgement. We applied this model to age-of-first-admission data for 4218 patients with ICD8/9 schizophrenia drawn from a state-wide psychiatric register. Both the raw age-of-first-admission distribution curve and the transformed data (‘estimated age-of-onset’) reinforce previous findings that (a) there is a wide range of age-of-onset and (b) the shapes of the curves differ between the sexes. Inspection of the mathematically derived distribution supports the proposition that (a) transformation for a lag effect produces a lower onset age and (b) including a variable length of lag produces a change in shape of the distribution. We propose that the mathematical transformation of age-of-first-admission data may have heuristic value, but requires further empirical data on which to base the assumptions of the model.