Land conversion and agricultural intensification reduce both on-farm and near-farm non-crop habitat for arthropod biodiversity, with potentially detrimental consequences for biological control of crop pests. The diversity of ichneumonid wasps, a large family of parasitoids, was sampled over three years and parasitism of two insect pests was measured in annual vegetable farms the following year; management practices were described and vegetation and landuse cover within 1.5Km were measured for each farm. Ichneumonidae species richness was positively associated with landscape-scale vegetative cover and field-scale crop diversity, but not with landscape-scale vegetation diversity; subfamily responses to both landscape vegetation classes and crop diversity varied. Species richness within Campopleginae and Cryptinae were positively associated with perennial vegetation and negatively associated with annual cropland, whereas diplazontine richness was positively associated with grasslands and negatively associated with freshwater. Baccharis shrubs and annual crop cover best explain the distribution of ichneumonid species, regardless of subfamily. Three ichneumonid and two braconid wasp species and tachinid flies parasitized Trichoplusia ni larvae, but major mortality was due to Hyposoter exiguae (Ichneumonidae) in May and Microplitis alaskensis (Braconidae) in September. Spring parasitism rates were positively associated with annual crop and grassland cover--an opposite pattern to the abundance and species richness of the Ichneumonidae samples. T. ni parasitism in fall was positively associated with grassland cover, pest control intensity, and decreasing tillage, not perennial vegetation cover. Parasitism of Brevicoryne brassicae was not associated with landscape vegetation or with farm management. Although greater on-farm crop diversity and perennial vegetation conservation in cropland-dominated landscapes were associated with greater richness of Ichneumonidae, neither perennial vegetation nor wasp richness was associated with high parasitism rates of sentinel caterpillars or aphids in the following year. These results suggest that elements of the landscape mosaic are needed to support diverse communities of natural enemies, but pest control services do not necessarily map on to patterns of arthropod diversity. Over the long term, a more diverse community may provide "insurance" against pest outbreaks if a dominant parasitoid is lost, but areas of overlap between biodiversity conservation and agricultural goals must be assessed critically.