OBJECTIVE A systems approach proposes that hospital adverse events (AE) represent a failure of the organization rather than the individual, and are more likely when sub-optimal working conditions occur. We analysed AE using a systems approach to (a) investigate the association between AE occurrence and "latent" risk factors, which included temporal, workload, skill mix and supervision issues, and (b) document interactions between clinically related risk factors. METHODS Prospective observational study. METHODS Regional paediatric intensive care unit. RESULTS Data from 730 consecutive nursing shifts over 12 months (816 patient episodes, crude mortality 7.2%) were analysed using logistic regression modelling. Two hundred eighty-four AE occurred during 220 of 730 (30%) shifts. There were 103 unit- and 181 patient-related AE; the latter occurred at a rate of 6.0 per 100 patient days. Factors associated with increased AE included day shift, average patient dependency, number of occupied beds and the presence of multiple, simultaneous management-related issues that compromised the supervisory ability of the nurse in charge. Factors associated with decreased AE included the presence of a senior nurse in charge, a high proportion of the shift filled by rostered permanent staff, and/or senior nurses, the number of admissions and discharges and, surprisingly, the presence of new junior doctors. Interaction effects were demonstrated between patient workload factors (bed occupancy and patient acuity) and also between nursing supervision factors (seniority of the nurse in charge and factors compromising the nurse's supervisory ability). CONCLUSIONS These findings may provide a framework for strategies to reduce AE occurrence.