A prospective study of women attending a surgical emergency department in an 8-month period showed that 117 (9%) had been battered. Offered in-patient treatment was accepted and completed by 58 women (the index group), while 59 declined or did not complete treatment (the drop-out group). Both groups were compared with age-matched controls. In the battered group there were more foreign-born women and more divorces than in the control group. The drop-out group differed from the index group either in being less severely injured and seeking only certification of injury for legal purposes, or in having more previous psychiatric morbidity. The consumption of somatic and psychiatric care during the preceding 10 years was significantly greater in the battered group than in the controls, probably due to the maltreatment and reluctance to report it spontaneously. Breaking of a battering pattern at an early stage requires awareness by surgeons, general practitioners and psychiatrists that recurrent injuries of unclear origin may be due to such violence.