Constitutional chromosome deletions result in wide ranging morbidity and often fatality. Information about risks and causes of death in these patients is important for counselling, and may illuminate the functions of the part of the chromosome deleted. There have been no cohort studies analysing mortality risks in persons with specific deletions compared with general population rates. We therefore conducted a cohort study following cause-specific mortality in 2,561 patients with autosomal chromosome deletions diagnosed by light microscopy or fluorescence in situ hybridisation at cytogenetic laboratories across Britain, 1965-2002. The commonest deletions were of 22q (544 patients), 15q (460) and 7q (210) and the least common 19q (0) and 20q (2). The prevalence of visible deletions of different chromosome arms was significantly inversely correlated with gene density of the arm (p < 0.001). Mortality was 11-fold raised in the cohort compared with the general population (standardised mortality ratio = 11.4 (95% confidence interval 10.0-12.8)), was significantly raised for each deletion with > or = 25 subjects in the study, and had a lower confidence limit > 10 for deletions of 1p, 1q, 3p, 4p, 5q and 22q. Overall, 29% of deaths were due to congenital anomalies; significantly raised mortality occurred also from many other causes, varying by chromosome and arm of deletion. The data imply that viability of foetuses with visible chromosome deletions may be inversely related to gene density, and that all visible and fluorescence in situ hybridisation-detectable deletions lead to much raised mortality, but the extent and causes of mortality vary according to the specific deletion.