The pyroclastic flow that issued from the Arenal summit crater on 28 August 1993 came from the collapse of the crater wall of the cone and the drainage of a lava pool. The 3-km-long pyroclastic flow, 2.2±0.8×10 6 m 3 in volume, was confined to narrow valleys (30-100 m wide). The thickness of the pyroclastic deposit ranged from 1 to 10 m, and its temperature was about 400°C, although single bombs were up to 1,000°C. The deposit is clast-supported, has a bimodal grain size distribution, and consists of an intimate mixture of finely pulverized rock ash, lapilli, small blocks, and cauliflower bread-crusted bombs, in which are set meter-size lava fragments and juvenile and non-juvenile angular blocks, and bombs up to 7 m in diameter. Large faceted blocks make up 50% of the total volume of the deposit. The cauliflower bombs have deep and intricate bread-crust texture and post-depositional vesiculation. It is proposed that the juvenile material was produced entirely from a lava pool, whereas faceted non-juvenile blocks come from the crater-wall collapse. The concentration and maximum diameter of cauliflower breadcrusted bombs increases significantly from the base (rockslide + pyroclastic flow) to the top (the pyroclastic flow) of the deposit. An ash cloud deposited accretionary lapilli in the proximal region (outside of the pyroclastic flow deposit), and very fine ash fell in the distal region (between 5 and 30 km). The accretionary lapilli deposit is derived from the fine, elutriated products of the flow as it moved. A turbulent overriding surge blew down the surrounding shrubbery in the flow direction. The pyroclastic flow from August 1993, similar to the flows of slid and rolled rather than being buoyed up by gas. They grooved, scratched, and polished the surfaces over which they swept, similar to a Merapi-type pyroclastic flow. However, the mechanism of the outpouring of a lava pool and the resulting flows composed of high-to moderate-vesiculated, cauliflower bread-crusted bombs and juvenile blocks have not been described before. Highfrequency earthquake swarms, followed by an increase in low-frequency volcanic events, preceded the 1975, 1993, and 2000 eruptions 2-4 months before. These pyroclastic flow events, therefore, may be triggered by internal expansion of the unstable cone in the upper part because of a slight change in the pressure of the magma column (gas content and/or effusive rate). This phenomenon has important short-term, volcanic hazard implications for touristic development of some parts on the flanks of the volcano.