Metabolic acidosis appears to be responsible for a variety of detrimental clinical and biological effects in uremic subjects, making its correction with dialysis mandatory. By utilizing the kinetic approach, pre- and post-dialysis target levels to be achieved by dialysis can be defined. Although a clear-cut bicarbonate concentration threshold, below which deleterious effects appear, has not yet been defined, it seems that a time-averaged concentration of 24 mmol/l is the gold standard for an optimal acid-base correction. However, apart from the bicarbonate levels, even the characteristics of the buffer used play a role in optimal, symptom-free acidosis correction. Currently, bicarbonate must be considered the base to be administered during treatment in the majority of patients. Due to the numerous factors affecting the bicarbonate mass balance (both patient related and technique related), the ideal dialytic dose for an individual patient is highly variable. Hence, the use of theoretical models may be very complex and impracticable for the individual patient on routine dialysis. Dedicated statistical models, obtained from in vivo data, are more useful and reliable. They allow us to tailor the bicarbonate prescriptions according to individual needs and differing dialysis methods.