The incidence of low back injuries among nursing personnel was studied in relation to frequency of lifting of patients. Information from the personnel records of a 770 bed facility was gathered for Licensed Practical Nurses (LPN, 143 workers), Nurse's Aides (NA, 252 workers), and Attendants (ATT, 20 workers) who worked for at least 12 consecutive months in a 40 month period beginning in January 1982. The subjects were divided into two groups composed of those whose jobs involved frequent and infrequent patient lifting. Back injuries included both those that did and did not involve lost time from work. Incidence density (ID) was defined as the number of injuries reported by a specific employee group divided by the sum of the person months of employment of group members. IDs for the frequent exposure group and the infrequent exposure group, respectively, were 0.00626 and 0.00083 (ID ratio 7.54) for LPN; ratio 1.07) for ATT; and 0.00743 and 0.00316 (ID ratio 2.35) for the three categories combined. Logistic regression analysis identified length of employment and frequency of patient lifting as significant variables associated with back injuries. Occupation was not a significant variable. Analysis of survival (length of time individuals continued to work before reporting a back injury) yielded survival rates of 89 percent for infrequent lifting and 79 percent for frequent lifting after the 1,215 days of the study (significance level 0.01). The authors conclude that it is lifting exposure, not occupation, that is an important factor in the incidence of back injury, and that it is imperative that additional ways for lifting and moving patients be found that do not involve actual patient lifting by personnel.