AbstractIn the production of integrated circuits (I.C.s), the ability to estimate the yield of working devices, or die, is economically important. This yield depends on the distribution of material and processing defects across the surface of the wafer of devices.A number of models have been proposed to explain the distribution of point defects on I.C.s. Most of these models are based on the Poisson distribution for describing the number of defects expected on a die. These models do not inherently allow for dependence between the number of defects expected on two adjacent die.This paper develops a model for generating an I.C. defect distribution that allows for dependence in the number of defects on die that are near one another. The correlation matrix and yields of observed data are found to compare favourably with theoretical results of the proposed dependent model. Simulation results are used to compare the new dependent model to a previously proposed independent model. These dependent-model simulations show defective die clustered near each other on the wafer, a property that has been observed on real wafers. Methods of parameter and yield estimation in the dependent model are discussed.