Background Inappropriate tibiofemoral joint contact loading during gait is thought to contribute to the development of osteoarthritis. Increased co-activation of agonist/antagonist pair of muscles during gait has commonly been observed in pathological populations and it is thought that this results in increased articular loading and subsequent risk of disease development. However, these hypotheses assume that there is a close relationship between muscle electromyography and force production, which is not necessarily the case. Methods This study investigated the relationship between different electromyography-based co-activation measures and articular loading during gait using an electromyography-driven model to estimate joint contact loads. Findings The results indicated that significant correlations do exist between selected electromyography-based activity measures and articular loading, but these are inconsistent and relatively low. However despite this, it was found that it may still be possible to use carefully selected measures of muscle activation in conjunction with external adduction moment measures to account for up to 50% of the variance in medial and lateral compartment loads. Interpretation The inconsistency in correlations between many electromyography-based co-activation measures and articular loading still highlights the danger of inferring joint contact loads during gait using these measures. These results suggest that some form of electromyography-driven modelling is required to estimate joint contact loads in the tibiofemoral joint. © 2013 Elsevier Ltd.