A statewide study to ascertain the number of ultrasound scans received by women in pregnancy, to identify the proportion having a scan at 16 to 20 weeks' gestation, and to establish where the scan at 16 to 20 weeks was performed was carried out between January, 1991 and June, 1992 in Victoria. Additional data were collected by midwives and entered on the perinatal morbidity statistics form routinely completed for all births. Of 52,319 women providing responses, 3.1% did not have a scan. Of the remaining 96.9% who had a scan, 73.5% were scanned at 16 to 20 weeks' gestation. Predictors of not having a scan were maternal birthplace and higher parity: previous perinatal death(s), and attendance at nonteaching hospitals predicted the opposite. Predictors of being scanned were location of hospital (country), maternal birthplace, higher parity and maternal age (< 20 years). Substantial differences in frequency and timing were found between hospitals attended. Factors associated with the pattern of scanning are not readily explicable in terms of risk of malformations or women's choices.