Mount Suswa is a low-lying Quaternary volcano in the Eastern Rift Valley of Kenya. It rises from the south-sloping floor of the Rift as an asymmetric, shield-shaped cone that covers an area of about 270 km2. The volcano consists of sodalite-bearing, phonolitic lava flows and subordinate proportions of pyroclastic rocks. The cone was built on a volcanic plateau composed of quartz-bearing, trachyte flood lavas (Plateau Trachyte Series; Baker 1958). The volcanic history of Mount Suswa can be divided into three major eruptive episodes: (1) formation of a primitive, shield-shaped volcano composed mainly of lava flows derived from central sources; (2) eruptions at the time of cauldron subsidence producing abundant pumice and thick lava flows, most of which issued from a ring-fracture zone outside, and concentric with, the caldera escarpment (‘ring-feeder’ lavas); (3) post-caldera lavas which partly filled the caldera and later built Ol Doinyo Nyukie volcano. Towards the end of the last eruptive episode an unusual collapse feature, in the form of a ‘ring graben’, was formed inside the older caldera. South of Mount Suswa a series of north-south linear faults transect the plateau basement of trachyte flood lavas. Near the southern periphery of Mount Suswa these faults die out, in some cases converging toward the centre of the volcano. Also in the south it is not always possible to distinguish between the quartz-bearing, trachyte flood lavas of the Rift floor and the sodalite-bearing flows from the central vents of Mount Suswa. (Sodalite-bearing flood lavas are known to be present.) The primitive volcano consists of lava flows, the earliest of which are the most voluminous. An unusual heterogeneous rock comprises the upper parts of the youngest primitive volcano flows. The rock consists principally of globules of lava moulded on to each other, each with a continuous glassy rim and a vesicular, crystalline core. The flows have been termed ‘globule-surface lavas' (Johnson 1968). Other heterogeneous glassy rocks on Mount Suswa resemble examples from the controversial 'froth flows’ described from various parts of Kenya (including Mount Suswa) by McCall (1965) and McCall & Bristow (1965). After a period of quiescence, a caldera was formed in the summit of the primitive volcano, the relationships between the caldera escarpment and the pumice and ring-feeder lavas on Mount Suswa are described in detail. These relationships are significant because they question the widely held assumption that, in calderas with thick pumice mantles, it is the rapid expulsion of pumice (producing a void in the magma chamber) that leads to the collapse of the magma chamber roof. This process, known widely as the ‘Krakataun’ mechanism (Williams 1941; after van Bemmelen 1929), is not applicable to Mount Suswa since the pumice and ring-feeder lavas mantle the caldera escarpment. Instead, as previously suggested by McCall (1963), a more likely process is that releases of pressure along the ring faults, formed during cauldron collapse, produce the explosive eruptions. In reviewing and discussing in detail the literature on calderas, it is concluded that many so-called ‘Krakatau-type’ calderas may have originated in the same way as the Mount Suswa caldera. Furthermore, it is emphasized that in the case of any one caldera great care must be taken in describing and interpreting the often ambiguous relations between cauldron subsidence and concomitant explosive eruptions. The post-caldera sequence of lavas on Mount Suswa is divided into two parts: an earlier group of generally non-porphyritic lavas; and a later group of distinctive, porphyritic lavas containing abundant anorthoclase phenocrysts. Most of the flows of the later group were erupted from a central vent in the southwest part of the caldera. They produced Ol Doinyo Nyukie volcano, at the summit of which is a pit crater. The second major collapse on Mount Suswa took place entirely within the older caldera and produced an annular trench, or ‘ring graben’. This unusual structure consists of two, more or less concentric, fault scarps bounding a steep-sided annular zone of subsidence. The ring graben truncates the pit crater of Ol Doinyo Nyukie and isolates a tilted, flat-topped, central island-block with a maximum diameter of 3.75 km. The island-block is inaccessible and its detailed structure is unknown. Consequently, the origin of the ring graben is still uncertain. However, three possible subsidence mechanisms are suggested. A fresh lava flow, similar to those of Ol Doinyo Nyukie volcano, partly covers the floor of the ring graben. This flow, and a similar one on the south flank of Mount Suswa, are the most recent eruptions of the volcano. Fumarolic activity persists at the present day.