Ovalipes catharus is a back-burrowing brachyuran which makes temporary burrows in soft sandy sediments. Like many burrowing crabs, O. catharus shows prolonged periods of sustained reverse ventilation. Well settled individuals were found to exclusively reverse-ventilate, regardless of burial state. However following activity, both buried and unburied animals showed periods of forward ventilation. Immediately following 15 minutes of exercise by swimming, crabs remaining unburied spent an average of 75.4 ± 14.2% of the time forward ventilating, while crabs that were allowed to burrow in sand spent only 32.5 ± 12.1% of the time in the forward mode. Previous studies suggest that buried crabs reverse the direction of ventilation in order to maintain a respiratory water stream by utilising super incumbent water. (Garstang, 1897b; Hartnoll, 1972; Taylor, 1984; McLay and Osborne, 1985; Otwell and Webb, 1990). Caine (1974) went further to suggest that interstitial oxygen tensions may be too low to sustain the oxygen demands of buried crabs. In laboratory studies the interstitial water in sediments inhabited by O. catharus, indeed, were found to be hypoxic. Values of interstitial Po2 were as low as 50% of that of the super incumbent water. Despite this, the present study has shown that both forward and reverse ventilation are utilised by O. catharus, regardless of burial state. When buried, Pio2 is maintained at near normoxic levels during forward ventilation by the utilisation of super incumbent water for ventilation. This is enabled by the formation and use of exostegal channels which provide a conduit for water flow from above the sediment to the primary inhalent openings, the Milne-Edwards apertures. In unburied O. catharus, branchial water flow patterns were found to be as described for other species (Hughes et al., 1969; McDonald et al., 1977). Water is drawn into the ventral hypobranchial space of the branchial chambers at the bases of the limbs and flows dorsally through the gills to the epibranchial space before passing anteriorly and being exhaled via ducts near the mouthparts. This flow is roughly countercurrent to the direction of haemolymph flow within the gills. When the direction of ventilation was reversed, this pattern of flow was disrupted. Anteriorly, water enters the prebranchial chambers, where the scaphognathites are situated before passing into the epibranchial space. Much of this water flows posteriorly and is exhaled via apertures between the fourth and fifth pereiopods. Some exhalent flow appears at the Milne-Edwards apertures suggesting some concurrent exchange may be possible. From the major sites of inhalation, the patterns of flow appear to be similar in buried crabs. However, the proportions of flow exiting each aperture were highly variable, suggesting an ability to regulate patterns of gill irrigation in this species. In unburied crabs, the mean oxygen extraction efficiency was significantly lower and the mean convection requirement (Vw/Mo2) was significantly higher during periods of reverse ventilation (Ew% = 23.7 ± 1.8%; Vw/Mo2 = 18.4 ± 1.7 ml µmol-1) when compared to adjacent periods of forward ventilation (Ew% = 32.9 ± 4.1%; Vw/Mo2 = 13.9 ± 1.9 ml µmol-1). In contrast, in buried crabs, values of Ew% and Vw/Mo2 were similar during adjacent periods of forward (Ew% = 53.3 ± 4.6%; Vw/Mo2 = 10.5 ± 1.4 ml µmol-1 ) and reverse ventilation (Ew% = 49.4 ± 2.9%; Vw/Mo2 = 9.7 ± 0.5 ml µmol-1). The lower values of Ew% and higher values of Vw/Mo2 recorded from unburied reverse ventilating crabs are presumably due to disruption of the counter current gill perfusion/irrigation relationship that exists when ventilating in forward mode. When buried, the improved efficiency of oxygen extraction and convection requirements seen during periods of reverse ventilation compared to unburied crabs utilising this mode, may be due to better irrigation of the gills resulting from regulation of the branchial water flow pathways, or to utilisation of an alternative site for gas exchange. The potential of the branchiostegal lining of the branchial chambers to fulfil such a role was examined. From vascular corrosion casts it was apparent that this region is well supplied with venous haemolymph which returns to the pericardium without being reoxygenated at the gills. This would create a large Po2 gradient across the cuticle of the branchiostegite which is relatively thin (10-20 µm), thus diffusion of oxygen across this barrier and into the haemolymph may be possible. Possible mechanisms of regulation of perfusion of the gills and branchiostegites are discussed. In unburied crabs, branchial chamber pressure (Pbranch) was similar in magnitude, but of opposite sign, in the two ventilatory modes. In buried animals, values of Pbranch were much greater than those recorded from unburied crabs, and Pbranch was significantly greater during periods of forward ventilation than reverse ventilation. The increased values of Pbranch in buried crabs indicate a greater resistance to ventilatory water flow, especially when ventilating in the forward direction. This increase in ventilatory resistance affected scaphognathite function in two main ways: Firstly, mean scaphognathite stroke volume (Vs) in buried crabs was reduced compared to unburied animals. In forward and reverse ventilating unburied animals, mean Vs was similar at 4.86 ± 0.14 ml beat-1kg-1 and 4.92 ± 0.15 ml beat-1kg-1, respectively, while in buried crabs, mean values of Vs were 3.63 ± 0.14 ml beat-1kg-1 and 4.11 ± 0.14 ml beat-1kg-1 in forward and reverse ventilation, respectively. Secondly, as a result of the higher values of Pbranch recorded from buried animals, ventilatory stroke work (Ws) and ventilatory power (Wr) were increased when buried, especially when utilising the forward mode. By converting ventilatory power terms to oxygen equivalents and using efficiency values from the literature (Wilkens et al., 1984), estimates of the energetic cost of ventilation (ie. the fraction of total Mo2 that is devoted to ventilation) were calculated for the four treatment groups at a given ventilatory flow (Vw = 0.6 1 kg-1 min-1). In unburied animals, the ventilatory power and oxygen requirements were similar in the two modes. However, as a result of the lower values of Ew% and Vw/Mo2 recorded during reverse ventilation, Mo2 was lower at a given Vw in this mode. Because of this, a higher proportion of total Mo2 (23.6%) was required for ventilation in the reverse mode, than the forward mode (12.1%). In buried crabs, Ew%, Vw/Mo2 and Mo2 were all similar at a given Vw. However, the increased ventilatory resistance seen in the forward mode, required an increased power output from the ventilatory muscles when generating flow, compared to that during reverse ventilation. This translates into an increased oxygen requirement of the ventilatory musculature. As a result, the estimated ventilator fraction during forward ventilation was higher (43.3%) than during reverse ventilation (20.0%). The predominant modes of ventilation in buried and unburied crabs following exercise reflect these differences. Unburied crabs primarily utilise forward ventilation while buried crabs utilise reverse ventilation. It is suggested that O.catharus modify their ventilatory behaviour depending on environmental factors and the internal physiological state of the animal, in order to reduce the overall energetic cost of ventilation. Potential mechanisms of control of ventilatory switching are discussed.