The behaviour of Phytophthora cinnamomi Rands was examined over a 2 year period on newly cleared and replanted eucalypt sites in the coastal forests of eastern Gippsland, rated as of low, moderate and high die-back hazard in relation to the drainage and disease characteristics of the original forest cover and surrounding trees. The fungus behaved like a saprophytic survivor. Maximum population levels, assessed as a 'population density index' (PDI), occurred within the soil influenced directly by the root mass. The PDI in adjacent bare soil was very low. The PDI depended, inter alia, on fungal pathogenesis, soil temperature and soil moisture. Mean maximum PDI values recorded respectively for the high, moderate and low hazard sites were in the ratio of about 8/4/1. This ratio was directly related to the depth of the clay pan beneath the surface. There was a marked seasonal variation in PDI, most pronounced on the high hazard site and least on the low hazard site. The minimum and maximum values were recorded in June (winter) and December-March (summer) respectively. Moderately heavy rainfall had little effect on the PDI in well-drained soils, even during midsummer when soil temperatures were optimum for fungal populations to increase. Fire affected PDI values only temporarily. PDI values in the surface soil were greater than at a depth of 75 cm, and the distribution of the fungus through the soil profile was influenced by soil texture. Before canopy closure occurred, PDI values were greater under eucalypt species tolerant to root rot than under sensitive species. Canopy closure reduced the seasonal fluctuation in soil temperature and, on sites where tree growth was vigorous, the reduction in PDI was striking. Addition of fertilizers to the soil had no direct effect on the PDI, the reduction observed being an indirect effect produced possibly by accelerated growth, increased transpiration and rapid canopy closure. These results suggest that root rot-sensitive eucalypt species growing on coastal sites with impeded drainage will prove vulnerable to die-back, and that any form of logging activity that reduces the amount of green cover on infected sites will aggravate root rot if the fungus is present. Consequently it will not be possible to manage these stands by conventional methods. On steeper, well-drained sites the disease should not prove to be a serious hazard except in exceptionally wet summer and autumn periods.