OBJECTIVE To examine training frequency's effect on torso rotation muscle strength. DESIGN The study followed a pretest-posttest randomized-group design. SETTING University laboratory. PATIENTS Subjects, 33 men (age 30 +/- 11yr) and 25 women (age 28 +/- 10yr) with no history of low back pain, volunteered to participate in the study and were tested for isometric (IM) torso rotation strength before (T1) and after (T2) 12 weeks of training. Measurements of maximal voluntary IM torso rotation torque (N.m) were made through a 108 degrees range of motion (54 degrees, 36 degrees, 18 degrees, 0 degree, -18 degrees, -36 degrees, -54 degrees). Subjects were stratified by peak torque at T1, and randomized to a nonexercising control group (C, n = 10), or groups that trained once a week (1x/wk, n = 16), twice a week (2x/wk, n = 17), or three times a week (3x/wk, n = 15); and all groups were similar in strength. INTERVENTIONS Training consisted of 8 to 12 repetitions of full range dynamic variable resistance exercise to volitional fatigue, for both left and right rotation. MAIN OUTCOME To determine the best training frequency for the development of torso rotation strength. RESULTS Relative improvements (average increase in strength gained at each angle) for the training groups were 4.9%, 16.3%, and 11.9% for the 1, 2, and 3x/wk groups, respectively. The 1x/wk group did not increase in IM torso rotation strength compared to the control group at any angle. Both the 2 and 3x/ wk groups increased their IM torso rotation strength compared to the control group at all but one angle. There were no significant differences in IM torso rotation strength between the groups that trained 2 or 3x/wk. During the training period, the 2 and 3x/wk groups increased their dynamic training load significantly more than the 1x/wk group. CONCLUSIONS Posttraining dynamic strength was not different between training frequencies of 2 and 3x/wk. Therefore, training the rotary torso muscles 2x/wk is recommended.