Values for carbon dioxide solubility (alpha CO2) and the apparent first dissociation constant (pKapp) of carbonic acid in rainbow trout plasma were measured at 5, 10 and 15 degrees C so as to eliminate the uncertainties with continued use of mammalian values extrapolated from the much higher temperatures of their determination. Estimates of pKapp were based on the in vivo measurement criteria most commonly used (i.e. whole blood pH, PCO2 and the CCO2 of true plasma separated from red cells at room temperature). Apparent pK varied inversely with pH, the dpKapp/dpH slopes at 10 and 15 degrees C (-0.075 and -0.080, respectively) being significantly elevated with respect to that at 5 degrees C (-0.004). At constant pH, dpKapp/dTemp varied between -0.0160 (pH 7.4) and -0.0208 (pH 8.0), both of which are higher than theoretically and experimentally based literature data on separated plasma. When we repeated our pKapp determinations (using identical methods) on rainbow trout separated plasma, we obtained dpKapp/dT slopes ranging from -0.009 to 0.0110, similar to all previous determinations. In attempts to account for the discrepancies between our whole blood and plasma based pKapp estimates, we found that the pH of whole blood was always lower than that of its isothermally separated true plasma (0.015 units lower at 15 degrees C) and that this difference became magnified at lower temperatures (0.033 units lower at 5 degrees C). Also, if cool blood was allowed to warm towards room temperature before and/or during anaerobic centrifugation for true plasma, CO2 was found to leave the red cells and result in a higher plasma total CO2 content relative to the amount contained in the original blood plasma (0.40 mM for a 15 degree C dT of separation). We conclude that use of pKapp values obtained from gasometric determinations on separated plasma is not appropriate for PCO2 or [HCO3-] calculation by the Henderson-Hasselbalch equation when the practice is to measure the whole blood pH and the total CO2 content of true plasma separated at temperatures other than that of the original blood plasma.