AbstractIt has been proposed that most cirques are source‐area depressions of large, deep‐seated rock‐slope failures. Yet the close relation between cirques and climate is convincing evidence of the dominance of glacial erosion, rather than rock‐slope failure, in mountain cirque development and distribution. Cirque floor altitudes have a lower limit that varies with snowfall by 1000 m or more between windward and leeward sides of mountain systems. Glaciation levels and equilibrium line altitudes implied by cirques vary in parallel with those for modern glaciers. Cirques are often found mainly on the poleward or leeward slopes of individual mountain ranges, as are modern small glaciers (because of solar radiation and wind effects on ablation and accumulation).Most rock‐slope failures (RSFs: rock slides, rock avalanches and gravitational deformations) do not involve the deep‐seated rotational movement that would produce a cirque form. Although some deep‐seated RSFs with arcuate head scars may be confused with cirques, identification as a glacial cirque is more confident as the floor is longer, wider and more gently sloping. Some scars from major RSFs may resemble poor or moderately developed cirques, but tend to have steeper floors, to be more scattered and closely related to geology, whereas glacial cirques develop on all rock types. Deep‐seated RSFs high on slopes can be associated with seismic shaking, but cirques develop without relation to seismicity. Degree of cirque development can be related to duration of exposure to glaciation. Often RSFs are found adjacent to cirques, or in glacial transfluences; only a proportion are well situated to develop into glacial cirques. Valley‐head cirques are continued down‐valley by glacial troughs. The ‘overdeepening’ (rock basins with reversed slopes) found in a large minority of cirques is not due to rock avalanching, fluvial or periglacial erosion.The RSF proposal should therefore be rejected in favour of the traditional glacial explanation, without any nivation stage being necessary. Rock‐slope failure is one of several possible ways of initiating hollows for glacier accumulation, as well as an ancillary process of cirque extension or widening through collapse of glacially oversteepened slopes.Headward extension of adjacent cirques on a ridge leads to displacement of the divide, sometimes by 2 km or more, lowering ridge and summit altitudes and producing the ‘glacial buzzsaw’ effect. Where a relatively lower snowline has led to cirque erosion on all sides of a mountain, cirque intersection lowers summits further. The buzzsaw hypothesis is not applicable, however, where remnants of a preglacial summit surface survive. © 2020 John Wiley & Sons, Ltd.