There is growing academic and policy interests in the factors that underpin the formation and the growth of clusters, especially for such 'hyped up' scientific and technological fields as the nanotechnologies. This paper analyses the determinants of scientific cluster growth (measured by the number of publications that emanate there from), distinguishing between structural effects (i.e. initial cluster size, scientific field composition and geographic location) on the one hand and its scientific variety, organizational diversity and degree of openness (in terms of collaboration with outside actors) on the other. Overall, scientific variety enhances clusters growth, but organizational diversity slows it down. However, patterns of growth are different in Asia, Europe and North America. It seems that cluster evolution is highly contingent on national systems of innovation and on the history of collaboration amongst local actors. Policy makers and cluster strategists must design specific policies by zone, and should not simply attempt to replicate best practices from one zone to another. Slow growth may reflect also 'elitist' strategies - those based on quality rather than on numbers