Reviews factors that affect the relative strength of top-down and bottom-up forces in food webs, and discusses methodological problems that colour our perceptions of the importance of these forces. Following a review of previous positions in the debate, the author examines consumer efficiency in multi-trophic level models, and considers feedbacks between consumers and resources, and between non-adjacent trophic levels. There are a small number of verbal and mathematical models which can accommodate synthetic, flexible views of changing, context-dependent roles of top-down and bottom-up forces in food webs. One problem lies in testing predictions when dynamic feedbacks may make outcomes highly time-dependent. Issues are illustrated, with emphasis on river food webs, concerning: positioning food webs along productivity gradients; quantifying trophic level biomass - the problem of omnivory; when to evaluate communities - the problem of incomplete recovery from disturbance; spatial circumscription of communities; and inferences from trophic level biomass accrual across regional productivity gradients. It is concluded that plants have primacy in food webs: in particular, their primary productivity is a fundamental control of higher trophic levels. -P.J.Jarvis