Measurement of unbound digoxin in presence of Fab fragments may be useful in management of overdoses. The analysis can be performed on serum directly or on ultrafiltrate of serum. The architecture of the immunoassay may influence the validity of results obtained using these two approaches. We tested this hypothesis by preparing serum mixtures containing various concentrations of digoxin and Digibind(R) and analyzed them by the immunoassays before and after ultrafiltration. Four samples collected from Digibind(R)-treated patients were also analyzed before and after ultrafiltration. The slopes and the y-intercepts of the measured versus the expected values for serum and its ultrafiltrate overlapped for the MEIA digoxin assay. I;or other three immunoassays tested (ACS:180, Stratus, and On-Line), either the slope or the intercept for measured versus the expected results for serum were significantly different (P<0.05) than those for ultrafiltrate. Following addition of digoxin and Digibind(R), differences in results for serum analyzed directly or after ultrafiltration were <0.50 ng/ml. Comparable samples from digoxin-overdosed patients treated with Digibind(R) had differences of >1.0 ng/ml. Previous claims reporting direct analysis of digoxin in presence of antidote but not having used patient samples for validation should be revisited. To date, analysis of serum ultrafiltrate by an immunoassay proven not to have matrix bias remains the most accurate approach in measuring unbound digoxin in presence of antidote. (C) 1999 Elsevier Science B.V. All rights reserved.