The symbiotic and saprophytic persistence of three unmarked Rhizobium leguminosarum biovar trifolii (Rlt) strains introduced into a field site in Iceland were followed. This site was free of clover cultivation and initially devoid of clover-nodulating rhizobia as tested by nodulation studies. Nodule occupancy by strains was identified based on their distinct ERIC-polymerase chain reaction (PCR) DNA fingerprint patterns. The survival and persistence of the individual strains in soil were monitored by the quantitative real-time PCR (qRT-PCR) assay, targeting the host-specific nodE gene. The most dominant strain in the nodule population, Rlt 20-15, showed relatively less saprophytic survival ability and maintained high numbers only in the presence of the appropriate host plant. Conversely, the minor nodule occupant, Rlt 32-28, persisted in soil at a relatively higher abundance both in the presence of its host legumes and in the presence of a non-host grass. The qRT-PCR assay was successfully applied to quantify rhizobial strains directly in soil without culturing or nodulation. However, the assay demonstrated less sensitivity compared with the plant infection most-probable-number (MPN) method for estimating the population size of rhizobia in soil. The quantitative detection limit of our qRT-PCR assays was 1 × 103 cells per gram of soil, as opposed to the MPN test which has a detection limit of 10 cells per gram of soil.