We developed 4 simple numerical models of plankton dynamics to explore how nutrient enrichment and habitat variability might influence the efficiency by which phytoplankton (P) production is transferred to growth of zooplankton (Z) consumers in coastal ecosystems. The 4 models range in complexity from 2 (P and Z) to 5 state variables (including detritus, nutrients, and 2 algal size-groups). The models employ generic equation formulations, which are generally well supported by empirical studies and are widely used in coastal ecosystem modeling. Simulation experiments revealed that trophic transfer efficiency (TTE = zooplankton growth per unit phytoplankton production) tends to be enhanced with increased variability of resources, particularly at low nutrient levels. Numerical and analytical studies also showed that, regardless of resource variability, these model formulations produce a trend of initial enhancement of trophic efficiency with increasing nutrient levels, followed by a marked reduction in efficiency beginning at moderately eutrophic conditions. This precipitous drop in trophic efficiency is attributable to a saturation of the ability of zooplankton to utilize the increased primary production associated with nutrient enrichment. Under these conditions, an increasing fraction of the primary production is shunted to microbial food chains and associated respiratory losses. The steepness of this reduction in trophic efficiency with nutrient enrichment is related to the strength of predation (or disease) control at upper trophic levels. Model formulations simulating more intense top-down control (i.e. increasing mortality rates with increasing Z abundance) resulted in sharper declines in TTE with increasing nutrients. We speculate that these model results may help to explain how observed reductions in relative fish yield (per unit primary production) in many shallow nutrient-enriched estuaries and lakes are related to interacting effects of cultural eutrophication and intense fisheries exploitation. Furthermore, we surmise that these relationships are robust characteristics of most existing aquatic ecosystem models.