OBJECTIVE--To investigate the risk factors for low back pain in hospital nurses, with particular emphasis on the role of specific nursing activities. METHODS--A cross sectional survey of 2405 nurses employed by a group of teaching hospitals was carried out. Self administered questionnaires were used to collect information about occupational activities, non-occupational risk factors for back symptoms, and history of low back pain. RESULTS--The overall response rate was 69%. Among 1616 women, the lifetime prevalence of back pain was 60% and the one year period prevalence 45%. 10% had been absent from work because of back pain for a cumulative period exceeding four weeks. Rates in men were generally similar to those in women. In women back pain during the previous 12 months was weakly associated with height, and was significantly more common in those who reported frequent non-musculoskeletal symptoms such as headache and low mood. After adjustment for height and non-musculoskeletal symptoms, significant associations were found with frequency of manually moving patients around on the bed, manually transferring patients between bed and chair, and manually lifting patients from the floor. In contrast, no clear increase in risk was found in relation to transfer of patients with canvas and poles, manually lifting patients in and out of the bath, or lifting patients with mechanical aids. Confirmation of these findings is now being sought in a prospective study of the same population. CONCLUSIONS--This study confirms that low back pain is highly prevalent among nurses and is associated with a high level of sickness absence. People who often report non-musculoskeletal symptoms were significantly more likely to report low back pain. Specific manual handling tasks were associated with an increased risk of back pain; however, no such association was found with mechanised patient transfers.