This prospective, observational, anonymous incident reporting study aimed to identify and correct factors leading to reduced patient safety in intensive care. An incident was any event which caused or had the potential to cause harm to the patient, but included problems in policy or procedure. Reports were discussed at monthly meetings. Of 390 incidents, 106 occasioned "actual" harm and 284 "potential" harm. There was one death, 86 severe complications and 88 complications of minor severity. Most were transient but the effects of 24 lasted up to a week. Most incidents affected cardiovascular and respiratory systems. Incident categories involved drugs, equipment, management or procedures. Incident causes were knowledge-based, rule-based, technical, slip/lapse, no error or unclassifiable. The study has identified some human and equipment performance problems in our intensive care unit. Correction of these should lead to a reduction in the future incidence of those events and hence an increased level of patient safety.