At the present time no reliable information is available about how midwives actually decide to record a person as Aboriginal or Torres Strait islander on the Perinatal Morbidity Statistics form. In 1993, Koori Health in the Department of Health and Community Services asked the Perinatal Data Collection Unit to undertake a study to determine how midwives make this decision. The study was conducted in seven country and city hospitals and two universities from August to December 1993. The hospitals ranged in size from 149 to 2359 births per year. During a routine education session the midwives were invited to take part in a focus group discussing 'Aboriginality'. Fifty-four midwives were involved in the study. Most mothers were not asked if they identified as 'Aboriginal or Torres Strait Islanders'. Midwives did ask women who they thought were 'Aboriginal' by appearance. Most midwives did not ask about 'Aboriginality' because they felt uncomfortable asking. They felt uncomfortable because they felt that both Aboriginal and non-Aboriginal women would feel ill at ease when asked if they identified as Aboriginal or Torres Strait Islanders. There were many reasons for this but they included negative perceptions of Aboriginal women. Some midwives made discriminatory comments. The conclusions of the study are that midwives need to know the reasons why they are requested to ask all women if they identify as Aboriginal or Torres Strait Islanders. Major barriers are often based on poor knowledge and understanding of definitions and issues. There was strong evidence of the existence of underlying beliefs and perceptions about Aboriginal people which 'inhibit' midwives from complying with the Data Collection Unit's explicit advice on the collection of this information.