The incidence of severe perineal tears acquired during vaginal childbirth varies considerably across hospitals but its use as a safety and quality indicator is in jeopardy because of problems associated with accurate detection and recording.

To understand and interpret time trends in the incidence of third‐ and fourth‐degree perineal tears among women giving birth vaginally in 18 public maternity hospitals in South Australia, taking into account individualised risk factors for each birth.

The risk‐adjusted probability of a third‐ and fourth‐degree tear was estimated for each of 65,598 singleton vaginal births (2002–2008), using a previously published regression model. The risk factors for each birth included maternal age; parity and ethnicity; assistance with instruments and episiotomy; shoulder dystocia; and infant birthweight. Plots of ‘excess’ tears were generated to help identify maternity services where the observed incidence of severe trauma differed from the expectation estimated from the risk profiles.

Three hospitals were identified at which there were systematically more tears than expected (given their risk profiles), and five hospitals were identified at which there were fewer tears. However, increased tearing at two hospitals coincided closely with improved advocacy for better detection and treatment of perineal tears (especially partial third‐degree tears).

Statistical process control methods provide a powerful means of investigating temporal variations in the incidence of outcomes like severe perineal tears. Third‐ or fourth‐ degree tears should be retained as a quality indicator of maternity services, but it is likely that many third‐degree tears currently go undetected.