Abstract. For short times and small clusters we check the validity of the 1935 Becker-Doring equation by Monte Carlo simulations in the square Ising model at and near its critical point; some discrepancies are found.Keywords: Clusters; Ising model; Relaxation.Droplet models for collective phenomena in fluids are quite old and have served a useful role in the prediction of phase transition properties. In some sense these droplets are the elementary excitations of fluids, just as phonons are elementary excitations in solids, and photons build up the radiation field. In the Debye theory for the specific heat of solids, we do not sum over all individual atoms, as Einstein did; instead we sum up the energy contribution from all phonons. Similarly, droplet models for fluids try to describe the fluid properties as a superposition of contributions from single droplets. Both the question of droplet-droplet interactions and the applicability of macroscopic droplet concepts to clusters containing one to ten atoms only have been controversial for many years, as has been even the definition itself of a droplet.More than half century ago, this journal published the Becker-Doring formula for droplet growth and decay [l]. Collective phenomena at liquid-gas (and analogous other) transitions have been described since 70 years by cluster or droplet models Here n, ( t ) is the actual time-dependent number of suitably defined clusters (droplets) containing s liquid molecules each, and N , is the equilibrium limit of that number; R , is the rate at which an s-cluster grows to size s + 1, and B, is the backward rate from size s + 1 to sizes. The current of net cluster growth from s to s + 1 isj, = R , n, -B , n,, , and must be zero in equilibrium: R, N , = B, N,,,. Thus the non-equilibrium current can be written asj, = R , N, (n,/N, -ns+l/Ns+r). The rate at which the actual cluster