To investigate nematode establishment and persistence, dauer juveniles (DJs) of Heterorhabditis bacteriophora were applied at 50 cm(-2) in different crops in June and July with conventional spraying equipment and 420 l water ha(-1). Application hardly had any effects on survival and infectivity. The number of DJs reaching the soil was assessed and the establishment and persistence recorded by baiting soil samples with larvae of the wax moth Galleria mellonella. The better the plant canopy was developed the fewer DJs reached the soil during application. Whereas in pasture 77% and in potatoes 78% of the applied nematodes reached the soil, in wheat and peas little less than 50%, in oil-seed rape only 5% and in lupine 6% were recorded. Between 50 and 60% of the soil samples contained H. bacteriophora a month after application with the exception of wheat (> 90%) and potatoes (< 5%) indicating that the number of nematodes reaching the soil during application had no influence on their establishment in the soil. Probably DJs can survive in the plant canopy and reach the soil later after application. The percentage of nematode-positive soil samples dropped considerably after tillage. In potatoes no nematodes were recovered after two months, which probably was also due to the intensive movement of the soil. Although nematodes are susceptible to freezing, temperatures below 0 degrees C during the winter did not extinguish the H. bacteriophora population. In field crops EPN usually persisted not much longer than one year. The longest persistence of H. bacteriophora was detected 23 months after release in beans followed in rotation by wheat with red clover as cover crop. In this field larvae of the pea weevil Sitona lineatus (Coleoptera: Curculionidae) were detected in soil samples and found infected with the released nematode population. In the laboratory the field soils were tested for persistence of H. bacteriophora at 8 degrees C and a half-life of 24.8 days was recorded in the absence of host insects and plants. Thus long-term persistence in the field was a result of recycling in host insects, which could not be detected in other crops than beans and clover. As H. bacteriophora seems to be restricted in its host potential, this species disappears after release once the host population is not available anymore.