We assessed soil fungal diversity and community structure at two sampling times (t1 = 47 days and t2 = 104 days of plant age) in pots associated with four maize cultivars, including two genetically modified (GM) cultivars by high-throughput pyrosequencing of the 18S rRNA gene using DNA and RNA templates. We detected no significant differences in soil fungal diversity and community structure associated with different plant cultivars. However, DNA-based analyses yielded lower fungal OTU richness as compared to RNA-based analyses. Clear differences in fungal community structure were also observed in relation to sampling time and the nucleic acid pool targeted (DNA versus RNA). The most abundant soil fungi, as recovered by DNA-based methods, did not necessary represent the most "active" fungi (as recovered via RNA). Interestingly, RNA-derived community compositions at t1 were highly similar to DNA-derived communities at t2, based on presence/absence measures of OTUs. We recovered large proportions of fungal sequences belonging to arbuscular mycorrhizal fungi and Basidiomycota, especially at the RNA level, suggesting that these important and potentially beneficial fungi are not affected by the plant cultivars nor by GM traits (Bt toxin production). Our results suggest that even though DNA- and RNA-derived soil fungal communities can be very different at a given time, RNA composition may have a predictive power of fungal community development through time.