The use of selective media to facilitate the isolation of Rhodococcus equi from environmental and clinical samples has aided studies of the ecology of R. equi and the epidemiology of disease caused by R. equi. Here, we compared the efficacy of two selective media (NANAT and modified CAZ-NB) for the recovery of six defined strains of R. equi and for the isolation and enumeration of both avirulent and virulent R. equi from 60 paired soil samples from horse farms using colony blotting and DNA hybridisation. No difference was found between the two media in the recoverability of defined strains of R. equi or the proportion of soil cultures positive for R. equi or virulent R. equi. NANAT medium was significantly less inhibitory of bacterial growth from soil culture compared to mCAZ-NB (P = 0.001), but there was no difference between the media in the number of R. equi colonies recovered. Soil cultured on mCAZ-NB medium yielded a significantly greater number of virulent R. equi colonies than NANAT (P = 0.03). The proportion of R. equi that were virulent in soil cultures on mCAZ-NB (32%) was more than three times that seen in cultures on NANAT (9%). Thus modified CAZ-NB appeared to be a better selective media for studies where the optimal recovery of virulent R. equi is required, such as in studies of the gastrointestinal carriage of virulent R. equi and of subclinically infected foals.