This study investigated the relationship between muscle morphology and surface electromyographic parameters (mean frequency, f(mean); and signal amplitude, RMS) during sustained static knee extension to exhaustion at 25% maximal voluntary contraction (MVC) and at 70% MVC. Twenty clinically healthy subjects participated. EMGs were registered from the quadriceps. Muscle forces during knee extension at a 90 degree joint angle were maintained at the respective levels throughout the measurement periods. A biopsy was obtained of the vastus lateralis muscle. The total time to exhaustion was normalized for each subject. By means of regression analysis, the intercept (i) (i.e. the unfatigued state) and the regression coefficient (k) were determined for each EMG parameter. The endurance time increased with decreasing force level. A significantly higher perception of fatigue was found at 25% MVC than at 70% MVC. From principal component analyses it was concluded that RMS-k at 25% MVC mainly correlated with the type 2 muscle fibre proportions (%), and at 70% MVC mainly with the areas of type 2 muscle fibre. At 25% MVC, f(mean)-k correlated with the areas of type 2A, 2B and 2C fibres, and at 70% MVC negatively with the proportion of type 2B and to some extent with areas of type 2A, 2B and 2C fibres. f(mean)-i at 25% MVC correlated with fibre type proportions (%); f(mean)-i at 70% MVC correlated with the areas of type 2A, 2B and 2C. The present study indicates relationships between surface EMG and muscle morphology, which is contrary to presented models of the EMG.