In this study we investigated whether the spatial distribution of surface electromyographic (EMG) amplitude can be used to describe the activation of muscle portions with different biomechanical actions. Ten healthy subjects performed isometric contractions aimed to selectively activate a number of forearm muscles or muscle subportions. Monopolar electromyographic signals were collected with an electrode grid of 128 electrodes placed on the proximal, dorsal portion of the forearm. The monopolar EMG amplitude [root mean square (RMS) value] distribution was calculated for each contraction, and high-amplitude channels were identified through an automatic procedure; the position of the EMG source was estimated with the barycenter of these channels. Each of the contractions tested was associated to a specific EMG amplitude distribution, whose location in space was consistent with the expected anatomical position of the main agonist muscle (or subportion). The position of each source was significantly different from the others in at least one direction (ANOVA; transversally to the forearm: P < 0.01, F = 125.92; longitudinally: P < 0.01, F = 35.83). With such an approach, we could distinguish the spatial position of EMG distributions related to the activation of contiguous muscles [e.g., extensor carpi ulnaris (ECU) and extensor digitorum communis (EDC)], different heads of the same muscle (i.e., extensor carpi radialis (ECR) brevis and longus) and different functional compartments (i.e., EDC, middle, and ring fingers). These findings are discussed in terms of how forces along a given direction can be produced by recruiting population of motor units clustered not only in specific muscles, but also in muscle sub-portions. In addition, this study supports the use of high-density EMG systems to characterize the activation of muscle subportions with different biomechanical actions.