OBJECTIVE: To determine the effect of oestrogen treatment on attenuating the growth of tall girls after adjusting for error in height prediction. DESIGN: Retrospective cohort study. PATIENTS: Tall girls assessed by Australian paediatric endocrinologists between 1959 and 1993. A total of 279 girls received oestrogen treatment (diethylstilboestrol or ethinyl oestradiol) and 367 girls were assessed but not treated. MEASUREMENTS: Estimated mature height (EMH) was calculated using radiographic assessment of bone age in adolescence. Final adult height was self-reported at follow-up. To control for error in the EMH predictions and their different distributions by treatment status, pairs of treated and untreated girls, matched on EMH within 1 cm, were selected for analysis. Covariate adjusted estimates of treatment effect (final height - EMH) were calculated. RESULTS In the sample of 108 matched pairs, the mean difference between the final height and EMH was -1.4 cm (SE 0.29) in the treated group and 1.1 cm (SE 0.23) in the untreated group, giving an unadjusted treatment effect of -2.5 cm (95% CI -3.2 to 1.8). A regression model based on 107 pairs of treated and untreated girls contained a significant interaction between bone age at treatment initiation and treatment, which estimated an approximately 1 cm per year decrease in treatment effect. The treatment effect was greatest in those commencing treatment at an early bone age and was significant if initiated before a bone age of 15 years. CONCLUSIONS: On average, oestrogen treatment resulted in an adult height that was less than predicted. Although treatment was more effective in the least mature girls, the mean height difference was relatively modest for most treated girls.