Domestic gardens associated with residential zones form a major component of vegetated land in towns and cities. Such gardens may play a vital role in maintaining biodiversity in urban areas, but variation in the abundance of organisms in gardens has been little explored. We report the results from a case study of 61 domestic gardens in the city of Sheffield, UK. Across 22 invertebrate groups, the median number of individuals recorded per garden was 49, 178, and 1012 in litter collections, pitfall and Malaise trap samples, respectively. Abundance was analysed by stepwise multiple regression and hierarchical tree analysis in relation to garden and landscape variables. The amount of variation explained in regression models ranged from 4 to 56%, for data based on pitfall and litter samples, and from 16 to 92% for data from Malaise traps. In total, 31 out of 36 explanatory variables entered into stepwise regression models, and 29 of them did so more than once. Although there was strong evidence only for approximately half of such relationships, in these cases the two methods of analysis corroborated one another. General correlates of invertebrate abundance were lacking, and likely reasons for inconsistencies in the relationships are discussed in the context of sampling and species biology. Correlates of the greatest significance occurred at both landscape (e.g. altitude) and garden scales (e.g. area of canopy vegetation). These factors were associated with species richness as well as abundance.