Topology provides a novel means to describe branching patterns and has not been applied to fungal colonies previously. For any branched structure, various topologies are possible, and these lie between two extremes, a herringbone pattern (main axis with primary laterals) and a dichotomous pattern (highly branched system). We applied topological methods to colonies of Peronospora viciae 48 h after inoculation of Pisum sativum leaves. The methods are based on two simulations, one developed for channel networks such as found in river systems and another for biological systems. Although not a true herringbone form, the Peronospora viciae colonies have a strong herringbone element within their growth pattern. All 25 colonies analysed fell into the random distribution according to the confidence limits calculated from simulations for biological systems. These confidence limits, however, represent the percentile distribution of all simulated networks, and only those structures with a perfect herringbone or dichotomous topology fall outside the range. The tendency of P. viciae colonies towards herringbone growth is reflected by the topological indices for altitude and external pathlength (a(obs)/E(a) and pe(obs)/E(pe), where a=altitude, pe=external pathlength, obs=observed for the P. viciae colonies and E=expected values for random growth), and the slope of the regression analysis for a(obs) and pe(obs). We consider this trend as significant because it was consistent for all but one of the colonies, and implies that growth can be envisaged as an intermediate between random and herringbone topology. It is proposed that initial herringbone growth may reflect a strategy that is aimed at overcoming host resistance, achieving rapid colonisation of infected tissue and maximising the potential for nutrient acquisition. This topology would also increase the likelihood of finding a compatible mating type for reproduction between heterothallic isolates.