The serum and urine levels of lidocaine and two active dealkylated metabolites, monoethylglycinexylidide (MEGX) and glycinexylidide (GX), were determined by HPLC in 33 cardiac patients receiving lidocaine for more than 1 day. Clinical assessment of nervous system toxicity was carried out at the time of blood drawing. The ratios in serum of MEGX to lidocaine and of GX to lidocaine were 0.36 +/- 0.26 (mean +/- SD) and 0.11 +/- 0.11. Lidocaine and MEGX binding to serum proteins from seven patients 2 days after their myocardial infarctions were 55.4 +/- 5.9% and 14.3 +/- 3.0%. After correction for this difference in protein binding, the MEGX/lidocaine ratio in serum water was 0.68 +/- 0.49. MEGX levels in serum water were 80% or more of the lidocaine levels in 11 of the 33 patients. GX binding was 5 +/- 4%. Even after correction for protein-binding differences, GX levels in serum water were low compared to lidocaine levels. The steady-state serum GX concentration normalized for lidocaine infusion rate declined with age. Of 27 patients without toxicity, six had serum lidocaine levels above 8 micrograms/ml. Five of six patients with toxicity had levels less than 8 micrograms/ml. The renal clearance of lidocaine, MEGX, and GX was less than creatinine clearance in most patients. We conclude that MEGX, but not GX, contributes to the pharmacologic activity of lidocaine therapy in a substantial fraction of patients. We also suggest that the concept of a single value for the upper limit of the therapeutic level of lidocaine in serum is an oversimplification because it does not take into account individual differences in drug-protein binding or accumulation of active metabolites.