Maximal voluntary isometric activations (MVIA) are frequently used as inputs for models attempting to predict muscle force and as normalization values in studies assessing muscle function. However, pain may adversely affect maximal muscle activation. The purpose of this study was to assess reliability of MVIA force and electromyographic (EMG) activity during prone isometric back extension in subjects with and without low back pain (LBP). A novel sub-maximal method using the percentages of the estimated mass of the head-arms-trunk (HAT) segment was also investigated. Repeated measures on 20 male volunteers divided into an LBP (n=10) and a control group (n=10) were made on 4 occasions. Force and EMG activity were recorded bilaterally from upper lumbar erector spinae (ULES), lower lumbar erector spinae (LLES), and biceps femoris (BF). Subjects exerted a maximal extension effort against a harness assembly that was attached to a force transducer. Submaximal exertions were also performed with an additional resistance of 100%, 110%, 120%, 130%, 140%, 150%, 160%, and 170% of HAT. Mean MVIA forces were significantly (p<or=0.05) lower in LBP vs. control. Intraclass correlation coefficients (ICC) for MVIA force, right and left ULES, and LLES EMG indicated high reliability in controls (R>0.90), but were significantly less in LBP (R=0.36-0.80). EMG of BF demonstrated excellent reliability across both groups (R>0.90). The resistance at 100% HAT demonstrated the highest reliability for LBP patients, whereas higher percentages of HAT showed either similar or higher reliability for controls. Force output and back EMG activity are less reliable with LBP individuals and should be taken into consideration when testing.