The 14C-DMO/3H-inulin method for pHi was critically assessed in intact Callinectes and found to be reliable provided adequate equilibration time and significant radiolabel excretion were taken into account. An unusually high ‘mean whole body pHi’ (7.54 at 20°C compared with a pHa of 7.80) was due to a highly alkaline fluid compartment (pHi = 8.23) in the carapace. At 20°C the pHi of the heart was 7.35 and skeletal muscle pHi was 7.30, and there were small but consistent differences in the pHi of different muscle types. The change in pHa with temperature was −0.0151 u°C−1 between 10 and 30°C, slightly less than the slope for the neutral pH of water (ΔpN/ΔT ≃ −0.0175 u °C−1). With data corrected to constant PiCoCo2 this was associated with a change in [HCO3−]a between 10 and 20°C (−0.13 mequivl−1°C−1, constant PaCoCo2) and a change in PaCoCo2 between 20 and 30°C (+0.13Torr°C−1, constant [HCO3−]a). The disturbing effect of relatively small PiCOCO2 changes on this pattern was demonstrated. ΔpHi/ΔT slopes for all tissues except carapace were not significantly different from pHa/ΔT but generally lower than ΔpN/ΔT. The slope for the. carapace was very flat and greatly influenced the ‘mean whole body pHi’ slope (−0.0062u°C−1). In haemolymph in vitro at constant Picoco2 ‘passive’ Δ[HCO3−]/ΔT (−0.17mequivI −1°C−1) was comparable to that in vivo between 10 and 20°C, independent of absolute PCOCO2. and directly related to total protein concentration. Haemolymph non-bicarbonate buffer value (β) was similarly related to protein, but increased with temperature. Crabs subjected to an acute 20→10°C shift showed initial overshoots of pHa and pHi associated with undershoot of PaCOCO2, all of which were corrected over 24 h as [HCO3−]a rose. During this period there was a significant net uptake