Most studies reporting more favourable biological features of screen-detected breast cancers compared with symptomatic or interval cancers include initial or prevalent screens and therefore may not indicate the real benefit of screening on breast cancer mortality. We conducted case–case comparisons within a cohort of eligible women (N=771 715) who were aged 50–69 between 1 January 1995 and 31 December 2003. A randomly selected sample of breast cancers (N=1848) diagnosed among these women were compared by detection method. Tumour characteristics of interval cancers (N=362) diagnosed after 6–24 months of a negative screen or symptomatic breast cancers (N=491) were compared with subsequent screen-detected breast cancers diagnosed within 6 months of a positive screen (N=995) using polytomous logistic regression. Tumours were evaluated for clinical presentation, histology and expression of hormone receptors. Women with symptomatic detected [odds ratio (OR)=7.48, 95% confidence interval (CI)=5.38–10.38] and interval cancers (OR=2.20, 95% CI=1.56–3.10) were more often diagnosed at stage III–IV versus I than women with rescreen-detected cancers. After adjusting for tumour size, women with symptomatic cancers had tumours of higher grade (OR=1.50, 95% CI=1.05–2.15) and mitotic score (OR=1.69, 95% CI=1.15–2.49) and women with interval cancers had tumours of higher mitotic score (OR=1.52, 95% CI=1.01–2.28) compared with women diagnosed at screening. Subsequent screen-detected cancers are not only detected at an earlier stage but are also less aggressive, leading to a better prognosis. As long-term mortality reduction for breast screening may depend on subsequent screens, our study indicates that mammography screening can be effective in women aged 50–69.