Demographic, behavioural, environmental, economic and obstetric history data from the Jamaican Perinatal Morbidity and Mortality Survey were examined to identify characteristics of women who do not attend for antenatal care, or present late instead of early for care, using multiple logistic regression. Non-attenders were more likely to be teenagers, unmarried, in unions of very short duration, smokers and women who felt that friends and relatives were not supportive. Multigravid non-attenders often had short inter-pregnancy intervals and included women who had experienced a post neonatal death. They were often drawn from deprived environments (lack of sanitation, water supplies). Late attenders shared features common to non-attenders (teenagers, unmarried, multigravid). Many however were self employed and did not fit the depressed profile of the non-attender. Most multigravidae who attended late had had previously uneventful pregnancies, including this one. Early attenders had little in common with non-attenders or late attenders. They were older, many had a secondary or tertiary education, were married and were generally middle class women. The group however included high risk multigravidae who had previous pregnancy complications or bad outcomes. Programmes aimed at reaching non-attenders must focus on the wider social and economic needs of these women and must give them a sense of their own power to effect change in their lives. Reaching the late attender will be more difficult and may be unnecessary with the possible exception of the teenager. She needs to be treated in a more sympathetic and non-judgmental way as this is often a high risk pregnancy. More fundamental changes require improved educational and employment opportunities for women as the best consumer is an educated consumer.