Copyright ©ERS Journals Ltd 1998 European Respiratory Journal

aa Worldwide, tuberculosis is the main infectious cause of death among adults, with most of these deaths occurring in developing countries [1–3]. Many tuberculosis-associ-ated deaths may be preventable by adequate chemother-apy. In developed countries, despite a decline in the overall tuberculosis-associated mortality, the risk of death of pati-ents with tuberculosis has increased, with an upward shift in the age distribution of tuberculosis cases [4–6]. In the Netherlands, the risk of tuberculosis infection has strongly decreased since the Second World War, leading to a low prevalence of infection among the young, although the prevalence of infection among the older cohorts is still high [7]. Consequently, as the incidence among the young age groups has decreased more strongly than that among the older ones, this cohort effect has shifted the age distri-bution of tuberculosis cases in the Netherlands as well. Recently, the total incidence of tuberculosis in the Neth-erlands has increased due to immigration from high-prev-alence countries [8–10]. Among the Dutch, tuberculosis in the younger age groups is increasingly concentrated in risk groups such as homeless and alcohol or drug addicts [7, 8]. It is conceivable that high-risk groups for tuber-culosis are increasingly difficult to reach by the health services with an adverse effect on treatment outcome. In addition, prognosis may be influenced by human immun-odeficiency virus (HIV) infection [11, 12]. The mainstay of tuberculosis control is treatment, in order to remove infectious sources from the population [2, 7, 13]. Treatment outcome is, therefore, extremely impor-tant in assessing performance of a tuberculosis control programme [14]. For the individual patients and their phy-sicians there is also a more immediate interest in treatment outcome, in particular in the probability of surviving the disease. In the present paper we estimate the size of excess mor-tality among tuberculosis patients and identify risk factors for tuberculosis-associated mortality for patients diagnos-ed in the Netherlands in the period 1993–1995, using sur-vival analysis