Elevated atmospheric CO₂ (eCO₂) is anticipated to have marked effects on soil microbial populations. While there is experimental evidence to support this view there are also studies where changes in microbial populations, such as the abundance of bacteria, are suggested but cannot be statistically established. We conducted this study to identify whether the sampling and sample treatment methods used could influence the results obtained using bacterial abundance as the variable of interest. We tested three different sampling methods and two different DNA extraction kits. The first because microbes are distributed heterogeneously in soil so the sampling procedure might be expected to influence the accuracy and precision of the population estimate and the second because the quantity and quality of DNA extracted influences the microbial analyses that can be performed and can introduce bias. Samples were taken from a long-running FACE experiment on grassland from under plants of Agrostis capillaris. We found that bacterial abundance was consistently lower under eCO₂ but we were only able to establish a statistical difference where a more intense sampling regime was used and bulking of the soil sample was avoided. A reduction in bacterial abundance is a consistent outcome in eCO₂ field experiments but the only other occasion where this reduction has been found to be significant was also where individual soil cores were analysed rather than the samples being bulked. We conclude that while there is extra work and cost attached to more detailed sampling this approach is highly desirable if we are to make robust conclusions about the impacts of eCO₂ on soil microbes.