Muscle blood flow can be constrained by external pressure or by internal pressure generated during isometric contractions. The loss of blood supply leads to metabolic fatigue but during submaximal contractions there are a number of ways in which the loss of performance can be compensated for, such as motor unit (MU) rotation and recruitment, or increasing firing frequency. We have examined recruitment and MU firing frequency during a sustained isometric contraction where blood flow was occluded. Twelve young men sat with their right leg extended and foot strapped into a dynamometer for measuring isometric dorsiflexions. Motor unit potential (MUP) amplitude and firing frequency were measured with an intramuscular concentric needle and the surface-representation (sMUP) determined using ‘spike triggered averaging’. Participants held a 25% maximal voluntary contraction (MVC) for 15 sec before a thigh cuff was inflated to occlude leg blood flow while subjects matched a target force of 25% MVC for as long as possible. Visual feedback was available throughout. MUs were sampled for 15 sec at the start of the contraction, half way through and at task failure. Subjects then rested, but with the circulation occluded, and made 15 sec voluntary efforts at 25% MVC at 60 sec and at 120 sec post-fatigue. Occlusion was then released, and 15 sec voluntary efforts at 25% MVC were measured after a further 60 sec. Data were analysed using repeated measures ANOVA and are reported as mean (s.e.m). Occluding the circulation to a fresh muscle had no immediate effect on MU size or firing rates. Half way through the contraction MU size had increased only slightly, but firing rates decreased from 10.7 (0.3) Hz to 8.4 (0.3) Hz. By task failure, larger MUs were recruited and firing rate (10.1 (0.5) Hz) was similar to those of fresh muscle. MU size remained elevated in the recovery period while occlusion was maintained and subjects reported difficulty holding the 25% MVC target for 15 sec. MU size rapidly returned to fresh values once the blood supply was restored. During the first half of the fatiguing contraction the results are consistent with peripherally-influenced reflex reduction in MUP firing frequency which may help to maintain electrical excitability and adapts to slow contractile properties of the fatiguing MUs, as suggested by Bigland-Ritchie et al for sustained maximal contractions (J Physiol, 379; 451-459). The increase in firing frequency during the second half of the contraction is indicative of the recruitment of larger, faster MUs. The novelty of these findings is that recruitment of new MUs is a relatively late event, there is no evidence of MU rotation as a strategy for maintaining force and the muscle remains in an apparent fatigued state if blood flow is occluded after the sustained effort.