A primary goal of the sociology of science is to understand the influence of social processes on the production of new scientific knowledge. There are three aspects of scientific knowledge that are generally considered important. First is its actual cognitive content. During the last 15 years the most emphasis has been placed on this dimension: understanding how social processes influence the construction of the actual cognitive content of science.1 Second is the foci of scientific attention, or how social variables influence which problems scientists decide to investigate. This has been a concern of sociologists of science since Merton's classic study of seventeenth-century English science.2 Third is the rate of scientific advance or how social, cultural and economic variables influence the amount of new knowledge produced. Although Merton's analysis in 1938 also dealt with the problem, in recent years relatively little attention has been paid to this aspect of the growth of scientific knowledge. We address this, using international data sets with nations as the unit of analysis. In an interesting report recently published in Science, Robert May has shown that there are great inequalities in the amount of science produced by various nations.3 Currently, the United States alone produces more than one third of the world's science, and the overwhelming majority is produced by a relatively small number of wealthy countries. It is also clear that which countries contribute the most to scientific knowledge is not a constant. It is generally accepted by historians of science that the centre of science was Italy in the sixteenth century, England in the second half of the seventeenth century, France around 1800, Germany around 1840, and the United States since 1920.4 Some people are concerned that in the future the United States may lose its current scientific preeminence. We will review theories which attempt to explain how national scientific productivity is influenced by social, cultural and economic forces, and present evidence enabling us to evaluate their adequacy empirically.