The Svalbard reindeer is the only mammalian herbivore in Adventdalen (78°N), Svalbard, where it has no natural predators. To test if herbivores in the absence of predators regulate standing crop to a level independent of productivity, which is one of the predictions of the “exploitation ecosystems” model, herbivore exclosures were set up in 1992 in Salix heath, Luzula heath, Cassiope heath, and Alopecurus meadow in Adventdalen. Standing crop of vascular plants was harvested and measured inside and outside the exclosures in 1994, when the reindeer population was at peak density (ca 5.4 animals km−2), and in 1996, when the reindeer density was about 30% lower (ca 3.7 animals km−2). Standing dead material was reduced by grazing in the Luzula heath in 1994. However, we found no effect of grazing, year, or interactions between grazing and year on live standing crop. Also contrary to the predictions from the model, differences in standing crop between vegetation types were highly significant. Mean biomass of plant material was lowest in the Alopecurus meadow (36 g m−2), two fold higher in the Luzula heath, and about threefold higher in the Salix heath and Cassiope heath, indicating that reindeer do not regulate standing crop to the same level on a local scale. The predictive power of the “exploitation ecosystems” model is low due to lack of recognition of the importance of plant chemistry, plant compensation ability, variation in forage availability during the year, parasites functioning as predators, and adverse weather conditions, which may cause density-independent variations in fecundity and mortality of reindeer.