Models for the steps of organisation in the origin of life are discussed with an emphasis on stability, and the possibilities of acquiring a diversity of functions. In particular, two basic models are described: that of simple self-replicating molecules, and that of autocatalytic self-reproduction, which is accomplished by a hypercyclic organisation. The latter may be exemplified by the RNA world. The view of a step-wise development with new functions successively incorporated and a high accuracy of the reproduction from the onset is criticised. Instead, we suggest that no clear systematic information is continued to the first cell before the start of protein synthesis. A non-selective manifold of self-replicating molecules and unsystematic protein production from the beginning could have caused a very large diversity from which functions that could stabilise the system by feedback loops could be selected. The only way to stabilise the protein synthesis and the genetic code would be to have feedback mechanisms so that the code actually produced the proteins that supported that very code. The code would then become frozen. As DNA would require control functions, it would not be used as a single information-carrier until protein synthesis had been established and the functions were available.