The invention relates to a P2P communication method for multi-subscriber networks, which is protected from deception, eavesdropping and hacking, and wherein the communication carried out in an interval is predominantly carried out in separate rooms, allocated to the P2P communication, and with separate reference data allocated to the P2P communication. At least part of the separate random reference data and/or random data is generated in at least one unit that participates in the P2P communication and is exchanged within the P2P communication in the form of relative data. The separate P2P communication is initiated with respect to at least one global random reference date valid for the time of the P2P communication, the random reference date being valid for a randomly determined time range and being stored in all units that carry out the P2P communications in a secret and non-deceivable manner.