A simple method for the detection of small populations of Pseudomonas fluorescens P.B8-1, containing the nptII gene of Tn5 as a unique marker, was applied to a Nyuzen paddy soil using cell extraction (indirect DNA extraction) and a "nested" polymerase chain reaction (PCR). This involved processing samples through a combination of a sucrose gradient centrifugation procedure to isolate bacterial cells, followed by cell lysis with proteinase K and CTAB (hexadecyltrimethyl ammonium bromide)-NaCl. This method allowed the extraction of DNA within about 6 h followed by amplification of DNA. The optimized "nested" PCR comprised a "2-step" PCR (45 cycles) using two 20-mer primers, followed by a "3-step" PCR (30 cycles) using two 26-mer primers which were internal to the first set. After the first PCR step was performed, the amplified DNA was detectable from the inoculated soil containing a minimum of 10⁵ cfu g^-1. However, the "nested" PCR procedure permitted the detection of amplified DNA fragments from inoculated non-sterile soils containing 1.3 × 10¹ cfu g^-1. The application of this detection strategy was tested by monitoring the survival of P. fluorescens P.B8-1 in a non-sterile paddy soil during a 53-day period. The P.B8-1 population decreased in soils maintained at either 25 or 10°C after inoculation. After 53 days, samples of soil maintained at 10°C contained 10² cfu g^-1 of P.B8-1 (as determined by selective plate count) and permitted amplification of DNA by the "nested" PCR. At the same time, P.B8-1 was not detected in soil maintained at 25°C by either method. The results obtained using this detection strategy suggest that it is highly applicable to monitoring the fate of genetically engineered microorganisms in natural paddy soils.