Reduction of migration of fecal coliforms (FC) and streptococci (FS) by limiting the leaching in effluent-irrigated soil was tested in lysimeters packed with quartz sand without or with added biosolids compost or with one of two clayey soils. The 200-L, 70-cm-deep lysimeters were either planted with a Eucalyptus camaldulensis or an Oroblanco citrus tree (in the sand only), or not planted. The Eucalyptus was irrigated with oxidation pond effluent (OPE) and the Oroblanco with mechanical-biological treatment plant effluent (MBTPE). The leaching fraction (LF) ranged from 0.2 to about 1.0, and the residence time (RT) from under 1 to 40 d. The Eucalyptus was also tested under intermittent leaching (RT 11-20 d) and deficit irrigation (without leaching for about 6 mo) regimes. Under MBTPE irrigation there was little or no leaching of FC and FS. Under OPE irrigation at LF 1 without a Eucalyptus there was little or no bacterial leaching at irrigation rates below 40 L d(-1) per lysimeter (RT > or = 0.8 d). Bacterial counts in the leachate were substantial in the presence of a Eucalyptus tree under LF 0.2 and intermittent leaching regimes, and when sand-packed unplanted lysimeters received OPE effluent at >45 L d(-1). Bacterial recovery peaked at LF 0.2, at up to 45% of the input level. At LF 1 (RT 0.6-2.8 d) and with intermittent leaching the recoveries were minute. Bacterial counts in the washout from the deficit-irrigated lysimeters were typical of nonpolluted soils. The bacterial concentration and recovery patterns in the leachate mostly matched the organic carbon (OC) load in the irrigation water, and its concentration and bioavailablity in the leachate. We related the leaching patterns of the fecal bacteria to their relative reproduction and die-off rates, and to the dependence of their regrowth on available carbon sources.