Agarose electrophoresis (Isopal, Beckman) and an immunoradiometric assay (IRMA) involving specific monoclonal antibodies (Ostase, Hybritech), two methods for the quantification of serum bone alkaline phosphatase (ALP, EC 3.1.3.1), a marker of osteoblastic activity, were compared in 293 patients: 79 with end-stage renal failure treated with hemodialysis and 214 with malignant disease. Overall correlation between the two methods was good (r = 0.92), except (a) for low values of bone ALP and (b) in some samples with high total liver ALP activity--both due to considerable cross-reactivity of the anti-bone ALP antibodies of the Ostase kit with liver ALP. This interference was not constant and was not evenly distributed across all concentrations of bone ALP. Low bone ALP determined with the IRMA (< or = 5 micrograms/L) was confirmed by electrophoresis (< or = 21 U/L), but bone ALP activity determined by electrophoresis to be low (< or = 21 U/L) was not correlated with the IRMA results. After standardizing our results by computing z-values for bone ALP, delta z (= zOstase - zIsopal) was significantly correlated with liver ALP activity (r = 0.73, P < 0.0001). We conclude that the IRMA for quantifying bone ALP is acceptable as a screening method. However, when high values for bone ALP are found with the Ostase method, confirmation by electrophoresis remains mandatory to rule out cross-reactivity with high amounts of liver ALP. For detecting low bone ALP activities, electrophoresis remains the method of choice.