With the species composition and/or functioning of many ecosystems currently changing due to anthropogenic drivers it is important to understand and, ideally, predict how changes in one part of the ecosystem will affect another. Here we assess if vegetation composition or soil chemistry best predicts the soil microbial community. The above and below-ground communities and soil chemical properties along a successional gradient from dwarf shrubland (moorland) to deciduous woodland (Betula dominated) were studied. The vegetation and soil chemistry were recorded and the soil microbial community (SMC) assessed using Phospholipid Fatty Acid Extraction (PLFA) and Multiplex Terminal Restriction Fragment Length Polymorphism (M-TRFLP). Vegetation composition and soil chemistry were used to predict the SMC using Co-Correspondence analysis and Canonical Correspondence Analysis and the predictive power of the two analyses compared. The vegetation composition predicted the soil microbial community at least as well as the soil chemical data. Removing rare plant species from the data set did not improve the predictive power of the vegetation data. The predictive power of the soil chemistry improved when only selected soil variables were used, but which soil variables gave the best prediction varied between the different soil microbial communities being studied (PLFA or bacterial/fungal/archaeal TRFLP). Vegetation composition may represent a more stable ‘summary’ of the effects of multiple drivers over time and may thus be a better predictor of the soil microbial community than one-off measurements of soil properties.