The purpose of this paper is to describe an integrated (meta)theoretical framework that I call anti-reductionist sociology: this entails a focus on agency, structure (the ‘conditions-of-action’), and social chance; micro–macro; and time–space. What follows allows only for a sketch of the framework; I have published fuller theoretical, methodological and policy-related accounts elsewhere. Anti-reductionist sociology is a sensitising theoretical perspective, not a body of substantive theory. This differentiation has an affinity with Mouzelis's (1993:684) distinction between methodological generalisations and substantive generalisations, the present paper being concerned with the former. In Cohen's terminology (1987:279–80) the approach outlined in the paper is ontologically flexible and in some sense eclectic: while explicitly excluding reductionist, essentialist, reified and teleological formulations, anti-reductionist sociology is a synthetic metatheoretical framework designed to encourage the development of substantive theories that are not necessarily tied to any particular paradigm.