In his ‘Simple model of herd behaviour’, Banerjee (1992) shows that – in a sequential game – if the first two players have chosen the same action, all subsequent players will ignore their own information and start a herd, an irreversible one. The points of strength of Banerjee’s model are its simplicity and the robustness of its results. Its weakness is that it is based on three tie-breaking assumptions, which according to Banerjee minimise herding probabilities. In this paper we analyse the role played by the tie-breaking assumptions in reaching the equilibrium. Even if the overall probability of herding does not change dramatically, the results obtained, which differ from Banerjee's are the following: players' strategies are parameter dependent; an incorrect herd could be reversed; a correct herd is irreversible. There are, in addition, some several cases where available information allows players to find out which action is correct, and so an irreversible correct herd starts.