By analysis of a previously described system consisting of three trophic levels (phytoplankton, zooplankton, and an invertebrate predator in a eutrophic lake), the generality of two hypotheses of food web theory was tested. (1) Top-down (predator-on-prey) impact weakens down the food web. (2) Time lags in the trophic level interactions decrease in duration from top to bottom of food webs. In the lake studied, a spring pulse of phytoplankton impacted with a delay the biomass of upper trophic levels - herbivorous zooplankton and the predatory water mite, Piona sp. Later in the season, the top-down control by Piona cascading down the food web became prevailing over bottom-up forces. Cross-correlations revealed that the top-down effect was not weakening downwards. Instead, the impact of zooplankton biomass on phytoplankton biomass was stronger than that of the invertebrate predator on zooplankton, despite the fact that the predator caused a decline in zooplankton biomass. Contrary to the theory, the time lags did not differ significantly between the impacts of trophic levels of a given pathway direction. However, they differed between bottom-up vs top-down pathways in support of the theory. Thus, some predictions of the food web theory may be restricted only to fish-impacted systems for which it was originally developed. An alternative explanation could be that the strength of top-down impact is associated with the width of consumers' feeding niches: the impact is stronger when niches are wider.