In many western Austronesian languages, the fact that an agentive argument lacks full control is morphologically marked on the verb. The formatives used for this purpose are often also found on stative predicates, and it has been suggested that limited-control predicates are stative-like in that they denote the result state of a given eventuality. Here we argue that limited-control predicates differ from both stative and dynamic predicates, and constitute a category of their own. Limited-control marking primarily pertains to agentivity and not to aspectual structure, and, importantly, is only used when control is at issue. With respect to the frequent overlaps with stative morphology, we argue that historically speaking, limited-control and stative marking have a common origin. While the current investigation does not include a full account of the historical developments leading to two synchronically separate categories (stative and limited-control), we provide evidence for the hypothesis that perception predicates had a major role to play in this development.(2) a. Ni-taip inang=ku taipang.RLS.UV-slice mother=1SG.GEN mango 'My mother sliced the mango.' b. Nikataipan inangku taipang. ni-ko-taip-an inang=ku taipang RLS.UV-POT-slice-UV2 mother=1SG.GEN mango 'My mother accidentally sliced the mango.'The major point of interest here is the fact that limited-control predicates as in (2)b overlap in their morphological marking with stative predicates. Thus, in (1)b the stative predicate is marked by the affix combination ni-ko--an, which is also found on the limited-control predicate in (2)b. As we will further illustrate in section 2, the degree to which stative and limited-control predicates morphologically overlap varies significantly across western Austronesian languages, ranging from almost total separation to total overlap.In cases where limited-control predicates differ morphologically from stative predicates, they are referred to by a wide range of terms in the Austronesianist literature, including accidental, involuntary, non-volitional, non-volitive, abilitative, potential, ability and involuntary action verbs, etc. Here we use the term potentive, following Rubino (1997) andHimmelmann (2004). In cases of total overlap, it is common practice not to distinguish between stative and limited-control predicates. Instead, the relevant forms are simply labelled stative and then different uses of these forms are distinguished. Kroeger (1990), for example, mentions 'possibility', 'attempted action', and 'non-volitional actions' among the uses of the stative affixes in Kimaragang (Sabah, Malaysia).