Objective. The purpose of this study was to investigate differences in consistency between maximum, preferred and a low level trunk performance. It was hypothesized that if maximum effort yielded consistent performances while submaximal efforts were less consistent, a consistency parameter could be used to detect lack of effort. Design. Survey study. Methods. Eleven healthy subjects performed lumbar dynamometry measurements (Isostation-B200) with maximum, preferred and low effort. A standard protocol was used and velocity and static torque were measured. Five subjects performed three measurements with low effort and trained the low effort level in between to determine the impact of training on consistency of submaximal performances. Coefficient of variations (CV), percentage differences and onset slopes were used to determine consistency. Results. Results show that maximum and preferred effort are equally consistent. Maximum and low effort and preferred and low effort show small significant differences in CV for some velocity and isometric torque parameters. No significant differences between the three performance levels were indicated for percentage differences and onset slopes. Training did not improve the consistency of the measurements. Conclusions. Small significant differences in CV were found between maximal and submaximal effort. However the combination with the relatively large standard deviations make it unlikely that this parameter can be used for a reliable discrimination.