OBJECTIVES: To investigate whether the size of the workforce (nurses, doctors and support staff) has an impact on the survival chances of critically ill patients both in the intensive care unit (ICU) and in the hospital. BACKGROUND: Investigations of intensive care outcomes suggest that some of the variation in patient survival rates might be related to staffing levels and workload, but the evidence is still equivocal. DATA: Information about patients, including the outcome of care (whether the patient lived or died) came from the Intensive Care National Audit & Research Centre (ICNARC) Case Mix Programme. An Audit Commission survey of ICUs conducted in 1998 gave information about staffing levels. The merged dataset had information on 65 ICUs and 38,168 patients. This is currently the best available dataset for testing the relationship between staffing and outcomes in UK ICUs. DESIGN: A cross-sectional, retrospective, risk adjusted observational study. METHODS: Multivariable, multilevel logistic regression. OUTCOME MEASURES: ICU and in-hospital mortality. RESULTS: After controlling for patient characteristics and workload we found that higher numbers of nurses per bed (odds ratio: 0.90, 95% confidence interval: [0.83, 0.97]) and higher numbers of consultants (0.85, [0.76, 0.95]) were associated with higher survival rates. Further exploration revealed that the number of nurses had the greatest impact on patients at high risk of death (0.98, [0.96, 0.99]) whereas the effect of medical staffing was unchanged across the range of patient acuity (1.00, [0.97, 1.03]). No relationship between patient outcomes and the number of support staff (administrative, clerical, technical and scientific staff) was found. Distinguishing between direct care and supernumerary nurses and restricting the analysis to patients who had been in the unit for more than 8h made little difference to the results. Separate analysis of in-unit and in-hospital survival showed that the clinical workforce in intensive care had a greater impact on ICU mortality than on hospital mortality which gives the study additional credibility. CONCLUSION: This study supports claims that the availability of medical and nursing staff is associated with the survival of critically ill patients and suggests that future studies should focus on the resources of the health care team. The results emphasise the urgent need for a prospective study of staffing levels and the organisation of care in ICUs.