The invention relates to a method for synchronization in networks, whereby the local time (t.sub.loc) which is valid at the particular node, is updated at different nodes. For that purpose, timing messages are regularly transmitted by a freely selectable superior node (N1; N3; N6) and only by a superior node to an inferior node N2, N3; N4-N6; N7), which receives the timing messages M1-M8) and analyzes said messages for updating the local time (t.sub.loc) thereof. A minimum propagation time (d.sub.min) is determined for a timing message (M1-M8) between an inferior node (N1;N3;N6) and a superior node (N2, N3; N4-N6; N7). When the inferior node (N2, N3; N4-N6; N7) receives a timing message (M1-M8), said inferior node extracts the local time of the superior node (N1; N3), which is contained in said timing message (M1-M8) and adds the minimum propagation time (d.sub.min) thereto, in order to generate a reference time (t.sub.comp,1-t.sub.comp,8). Said reference time (t.sub.comp,1-t.sub.comp,8) is then compared with the proper local time (t.sub.loc). If the reference time is retarded in relation to the proper local time (t.sub.loc), said proper local time (t.sub.loc) is not updated. If said reference time is advanced in relation to the proper local time (t.sub.loc).