Following publication of On the Origin of Species, biologists concentrated on and resolved the mechanisms of adaptation and speciation, but largely ignored extinction. Thus, extinction remained essentially a discipline of palaeontology. Adequate language is not available to describe extinction phenomena because they must be discussed in the passive voice, wherein populations simply ‘go extinct’ without reference to process, specifics, effects, or causality. Extinction is also described typically in terms of its dynamics (including rate or risk), and although correlative variables enhance our ability to predict extinction, they do not necessarily enable an understanding of process. Yet background extinction, like evolution, is a process requiring a functional explanation, without which it is impossible to formulate mechanisms. We define the mechanism of background extinction as a typically long-term, multi-generational loss of reproductive fitness. This simple concept has received little credence because of a perception that excess generation of progeny ensures population sustainability, and perhaps the misconception that the loss of reproductive fitness somehow constitutes selection against reproduction itself. During environmental shifts, reproductive fitness is compromised when biotic or abiotic extremes consistently exceed existing norms of reaction. Subsequent selection will now favour individual survival over reproductive fitness, initiating long-term negative selection pressure and population decline. Background extinction consists typically of two intergrading phases: habitat attenuation and habitat dissolution. These processes generate the relict populations that characterize many species undergoing background extinction. © 2011 The Linnean Society of London, Biological Journal of the Linnean Society, 2012, 105, 255–268.