Sodium was determined by flame photometry and by direct potentiometry in 56 serum or plasma samples from 24 patients with multiple myeloma or macroglobulinemia. We observed differences between the two techniques as large as 17 mmol/L (12%). The flame-photometric values decreased relative to the direct-potentiometric values as protein increased or water content decreased. Moreover, the two sodium measurements could not be interconverted simply on the basis of correcting for protein or water content. There was significantly lower residual variance (p<0.005) when the direct-potentiometric sodium values were compared with the osmolality (corrected for the influence of glucose and urea nitrogen) than when the flame-photometric values for sodium were so compared. We conclude that direct potentiometric measurements of sodium in patients with multiple myeloma gives clinically relevant results but flame photometry does not. Clearly, the method by which sodium is measured in patients with multiple myeloma must be considered if results are to be interpreted correctly.