A series of miscible-displacement column experiments were conducted under saturated flow conditions to systematically investigate the influence of physical and biological complexity on bacterial activity and fate in the presence and absence of a non-sorbing growth substrate, salicylate. Bacterial elution was monitored for three different systems; System I--a sterilized, inoculated, well-sorted sand, System II--a sterilized, inoculated, heterogeneous loamy sand (Hayhook), and System III--two different unsterilized loamy sands (Hayhook and Vinton) each with their associated indigenous microbial community. Results show that System I behaved ideally with respect to both cell and substrate transport, wherein: (1) growth occurred in response to substrate addition, (2) cell elution increased in response to the substrate pulse, and (3) breakthrough curves were reproducible for both substrate and cell elution. In contrast, System II showed ideal behavior with respect to substrate transport but showed variable behavior for cell transport. Further, there was no measurable growth in response to substrate addition and no increase in cell elution during the salicylate pulse. System III exhibited non-ideal behavior for both substrate and cell transport. Of particular interest is the fact that the indigenous communities of the two soils behaved differently. Specifically, for the Hayhook soil, an increased elution response was observed for the heterotrophic population while the salicylate-degrading community was preferentially retained in the column. In contrast for the Vinton soil, the substrate pulse did not elicit an elution response from either the heterotrophic or salicylate-degrading community from the culturable, indigenous Vinton microorganisms. For Systems II and III, the observed variability appears to be associated with the biological component of the system, since sterile controls were reproducible. This type of systematic study is critical for understanding cell and substrate transport behavior in complex, heterogeneous systems, and illustrates the potential uncertainty associated with measurements in such systems.