Crystal nucleation is a much-studied phenomenon, yet the rate at which it occurs remains dif®cult to predict. Small crystal nuclei form spontaneously in supersaturated solutions, but unless their size exceeds a critical valueÐthe so-called critical nucleusÐthey will re-dissolve rather than grow. It is this rate-limiting step that has proved dif®cult to probe experimentally. The crystal nucleation rate depends on P crit , the (very small) probability that a critical nucleus forms spontaneously, and on a kinetic factor (k) that measures the rate at which critical nuclei subsequently grow. Given the absence of a priori knowledge of either quantity, classical nucleation theory 1 is commonly used to analyse crystal nucleation experiments, with the unconstrained parameters adjusted to ®t the observations. This approach yields no`®rst principles' prediction of absolute nucleation rates. Here we approach the problem from a different angle, simulating the nucleation process in a suspension of hard colloidal spheres, to obtain quantitative numerical predictions of the crystal nucleation rate. We ®nd large discrepancies between the computed nucleation rates and those deduced from experiments 2±4 : the best experimental estimates of P crit seem to be too large by several orders of magnitude. The probability (per particle) that a spontaneous¯uctuation will result in the formation of a critical nucleus depends exponentially on the free energy DG crit that is required to form such a nucleus:where T is the absolute temperature and k B is Boltzmann's constant. According to classical nucleation theory (CNT), the total free energy of a crystallite that forms in a supersaturated solution contains two terms: the ®rst is a`bulk' term that expresses the fact that the solid is more stable than the supersaturated¯uidÐthis term is negative and proportional to the volume of the crystallite. The second is à surface' term that takes into account the free-energy cost of creating a solid±liquid interface. This term is positive and proportional to the surface area of the crystallite. According to CNT, the total (Gibbs) free-energy cost to form a spherical crystallite with radius R is DG 4 3 pR 3 r S Dm 4pR 2 g 2 where r S is the number-density of the solid, Dm (,0) is the difference in chemical potential of the solid and the liquid, and g is the solid±liquid interfacial free energy density. The function DG goes through a maximum at R 2g= r S jDmj and the height of the nucleation barrier is:The crystal-nucleation rate per unit volume, I, is the product of P crit and the kinetic prefactor k:The CNT expression for the nucleation rate then becomes: (6). A prerequisite for the calculation of the nucleation barrier is the choice of à reaction coordinate' that measures the progress from liquid to solid. As our reaction coordinate we use n, the number of particles that constitute the largest solid-like cluster in the system. A criterion based on that in ref. 8 was used to identify which particles are solid-like. If two solid-like particles are less t...