OBJECTIVE Unplanned extubation (UE) is an important paediatric intensive care unit (PICU) quality indicator. Studies on UE have been modest in size, with accurate UE rate calculation potentially hampered by ventilation episodes recorded in calendar days. We wished to document UE rates, outcomes, associated factors and quantify error when calendar days rather than exact timings are used. METHODS We recorded prospectively all UE episodes and potential associated factors in our 20-bed PICU for 12,533 admissions (2000-2013). Ventilation episodes were recorded to the minute, with non-invasive and tracheostomy ventilation excluded. Analysis utilised multilevel mixed-effects Poisson regression, adjusting for multiple ventilation episodes in the same patient. RESULTS Overall, 243 UEs occurred within 14,141 ventilation episodes (31,564 intubated days), giving a UE rate of 0.77 (95% CI 0.67-0.87) episodes per 100 intubated days. If calendar ventilation days were used, the yearly UE rate was underestimated by 27-35%. UE rates decreased with time, by approximately 0.05/100 intubated days each year. Associations with UE incidence rate included patient age, source of admission, disease severity and diagnostic category, with nasal tubes decreasing the risk. Although UE versus planned extubation was associated with a higher re-intubation rate (43 versus 8%) and longer median PICU stay (4.6 versus 2.6 days, p < 0.001), mortality between the two groups did not differ (3.0 versus 5.1%, p = 0.18). CONCLUSIONS This study provides contemporaneous UE rates for benchmarking. Recording ventilation in calendar days underestimates UE rate. Several factors associated with UE may serve as a focus of quality improvement.