The purpose of this investigation was to examine the test-retest reliability of non-normalized (absolute or raw units of mV s) and normalized (percentage ratio) scores using integrated electromyography (IEMG) in voluntary maximal isometric contractions. Bipolar surface electrodes on the vastus medialis (VM) and vastus lateralis (VL) muscles were used to record IEMG (mV s) on two occasions, 2-8 days apart. While positioned supine, 20 healthy young women (mean age 24 ± 2 yr) performed: (a) maximal voluntary unilateral quadriceps setting (QS) (i.e. isometric knee extension with the knee in 0°) with the ankle joint in a neutral position and the ankle musculature relaxed, and (b) maximal unilateral QS with the ankle joint in maximal dorsiflexion and the ankle dorsiflexors maximally contracted (QSD). The QS exercise was used to normalize IEMG (QSD : QS ratio for each of the two muscle heads) and the VM muscle was also used to normalize IEMG to the other head of the vasti (VL : VM ratio for each exercise). Reliability coefficients were excellent for test-retest reliability of non-normalized IEMG (mV s) and for IEMG normalized to another muscle: (intraclass correlation coefficients (ICCs) >0.86). However, IEMG normalized to another exercise was characterized by poor reliability (ICCs <0.34), even when determined as the reliability of data averaged over two occasions. Regardless of whether absolute or normalized IEMG was used, 95% confidence intervals were wide, suggesting that precise, interday prediction of an individual's performance using the present protocol is questionnable.