P>Objectives To describe clinical characteristics of severe acute respiratory syndrome (SARS) patients in a hospital in Tianjin, China, thereby comparing probable and suspected cases; to study risk factors associated with the death of cases; to describe the implementation of preventive interventions during the hospital outbreak. Methods Physical and haematological information was obtained from clinical records. White blood cell counts, and percentages of neutrophilic granulocytes and neutrophilic lymphocytes were measured. The service department of the hospital provided information about daily use of protective materials. Differences in clinical symptoms between probable and suspected SARS cases were tested by Fisher's exact test. Non-linear mixed modelling was used to test for differences between the haematological patterns for probable and suspected cases. Risk factors for dying among probable SARS cases were tested by logistic regression. Results The SARS outbreak started with a patient from Beijing on 15 April 2003, and spread quickly among the healthcare workers and in-patients in the hospital. In total 90 probable and 21 suspected cases were reported, with 17 deaths among them (case fatality rate 15%). Haematological patterns were significantly different between probable and suspected cases, whereas the percentages with certain clinical symptoms showed no apparent difference. Death of probable SARS cases was only significantly associated with high age and use of a respiratory machine (mainly for the most severe cases), whereas e.g. co-morbidity and steroid treatment showed no impact in multivariate analysis. Stringent control measures, including distribution of huge numbers of protective materials, started on 20 April, which soon lead to a strong decrease in the incidence of new cases. After the last SARS case left the hospital on 6 June, protective materials were dispensed at substantially lower rates, but not to zero, as was the case during the first days of the outbreak. Conclusion The working definition of probable SARS used during the epidemic appeared to have been fairly accurate. Many valuable lessons were learned regarding prevention of hospital spread of infection, especially the need to have sufficient protective supplies available and to implement these rigidly and at an early stage of an (threatening) epidemic.