Background/Question/Methods The main thesis of this paper is that our perception of ecosystems is is still strongly determined by a physiognomic perspective on nature. Especially if it comes to assessing the functioning or "dysfunctioning" of ecosystems, this aspect – although often implicitly – is as important as describing flows of energy and matter, which are "officially" seen as the decisive hallmark of ecosystem ecology. As physiognomic approaches have been considered as problematic both with respect to their philosophical assumptions as well as for methodological reasons, the question is as to whether this persistence is a matter of concern for ecology. Physiognomic approaches date back to Alexander von Humboldt, describing vegetation by means of broad-scale morphological characters, abstracting from specific species identities. In plant geography and later vegetation science, this approach formed the basis for the classification of plant formations. For several reasons, physiognomic approaches became unpopular in ecology, or in some of its forms were even considered as bad science. Nevertheless, physiognomic approaches persisted within ecology and, as I will demonstrate, physiognomy implicitly even constitutes one of the most important and common criteria in defining ecosystem boundaries and assessing ecosystem functioning. Results/Conclusions Physiognomic perspectives on ecosystems can be communicated very easily to different audiences. In fact, the physiognomic approach may allow for an easy and rapid assessment of ecosystem boundaries and even ecosystem functioning (in some of its meanings). Ecosystems perceived under this perspective do, however, not necessary coincide with those defined and delimited by other criteria, e.g. by emphasizing ecosystem services, or specific species compositions. Also, there are problems to clearly operationalize classifications based on physiognomy. To explore their potentials and limits, physiognomic approaches, as the basis of one (of several) useful ecosystem definitions, should be made more explicit and made a topic of research. Such research must focus on the empirical correlations between physiognomic delimitations of ecosystems and other types of ecosystems, as well as on the limits of building classifications on physiognomic criteria. For what purposes is a physiognomic characterization of ecosystems useful and for which not? Also, it is important to clarify the hidden assumptions behind this perspective, e.g., the connections of physiognomic classifications to aesthetics, which formed one of the starting points of Humboldt's idea.