Abstract The phenomenon of Partial Control (PC; cf. Landau in Elements of control: structure and meaning in infinitival constructions, Kluwer, Dordrecht, 2000) has received a great deal of attention in recent literature, with two general approaches: while some authors take PC to be a core phenomenon that should be captured by control theory (e.g., Landau 2000 et seq.; Pearson in The sense of self: topics in the semantics of de se expressions. Ph.D. dissertation, 2013; NLLT 34:691–738, 2016), others deny the relevance of PC to control theory by treating it as an instance of exhaustive control (EC) with a covert comitative (Hornstein, in: Hendrick (ed) Minimalist syntax, Blackwell, Oxford, pp 681, 2003; Boeckx et al. in Control as movement, Cambridge University Press, Cambridge, 2010). In this paper, we take a closer look at what German can contribute to this debate. Since PC-data are known to be subtle in terms of acceptability, we submitted the phenomenon to experimental investigation. The results of our study clearly show that both the matrix and the embedded predicate are relevant for the availability of a PC-reading in German: if the embedded predicate does not license comitatives, the matrix predicate must be attitudinal (a PC-predicate in Landau’s 2000 terminology), while if the matrix predicate is non-attitudinal (Landau’s EC-type), the embedded predicate must license comitatives. We thus conclude that two mechanisms exist in German to derive a PC-reading (i.e., singular controller with an embedded collective predicate): (i) true PC via, e.g., extension in the sense of Pearson (2013), and (ii) fake PC, which is, in fact, exhaustive control with a covert comitative (Hornstein 2003; Sheehan in Camb Occas Pap Linguist 6:1–47, 2012; in: Lahousse, Marzo (eds) Romance languages and linguistic theory 2012: selected papers from ‘Going Romance’ Leuven 2012, pp 181–198, 2014). These conclusions have an impact on the discussion of obligatory control structures insofar as control theory must account for the inclusion relation that holds in cases of true PC. Based on our results, we propose that a multi-factor approach such as argued for by Pearson (2013, 2016) or Landau (A two-tiered theory of control. Linguistic inquiry monographs, vol 71, MIT Press, Cambridge, 2015) is most adequate to capture the variation found in instances of true PC. However, our study also shows that the peripheral nature of PC requires experimental methods in order to investigate its status as a grammatical phenomenon in a language, and, furthermore, that any study focusing on PC must ensure that its discussion is based on true instances of PC, given that fake PC is a real option in languages.