In a multi-sources environment, the acoustical information from the different sources is mixed when reaching the listener’s ears. The auditory system is then able to organize sound into auditory objects using grouping mechanisms based on many cues such as harmonicity, common onset and offset, common modulation, etc. The way in which information is grouped has already been shown to significantly influence the perceived loudness either within (McAdams et al. 1998) or across (Scharf, 1969; Grimault et al. 2007) auditory channels. As a first observation, grouping all parts of the mixture into a single auditory object gives rise to a global loudness. Moreover, when different parts of the mixture are segregated into distinct auditory streams, each stream both results in a particular loudness (partial loudness) and contributes to a global loudness. Two main questions then arise: Does global loudness depend on the perceptual organization of the mixture (one or several streams)? How the partial loudness of different auditory streams are combined into one global loudness percept? In a recent experiment, frequency modulation was used to either separate the odd (A) and even (A’) components of six-component harmonic complexes or group all components together (B). First, A and A’ were equalized in loudness for each listener. Then, the partial loudness of A and A’ and the global loudness of A+A’ and B have been assessed. The results will be discussed, in order to provide possible answers to the questions aforementioned.