To test the hypothesis that lidocaine passage into the central nervous system is a function of free rather than total lidocaine concentration, we examined the effect of serum protein binding on the distribution of lidocaine into brain and cerebrospinal fluid (CSF) in dogs. Six of the 13 dogs studied were pretreated with rifampin to induce a 4-fold increase (P = 0.019) in serum concentration of alpha 1-acid glycoprotein, which is a major binding protein for lidocaine. The dogs were anesthetized and prepared surgically to obtain samples of cortical brain tissue or CSF from the cisterna magna. Lidocaine at a dose of 3 mg/kg was infused intravenously over 15 s. Arterial blood and brain cortex or CSF were sampled serially during a 60-min interval and analyzed for lidocaine content. Unbound or free fraction of lidocaine in serum was measured by equilibrium dialysis. Rifampin pretreatment led to a significant decrease in average serum free fraction of lidocaine, from 0.24 +/- 0.08 to 0.080 +/- 0.030 (P less than 0.001). Total lidocaine concentration in serum was higher, but free lidocaine concentration was lower in the rifampin group. Equilibration of lidocaine between serum and brain or CSF was reached by 10 min after lidocaine administration. Rifampin-pretreated dogs had consistently lower partition ratios of lidocaine between brain and serum or between CSF and serum. A strong and positive correlation between time-averaged brain to serum or CSF to serum ratios and serum free fractions were observed (r = 0.92, P less than 0.001 for brain; r = 0.90, P less than 0.01 for CSF).(ABSTRACT TRUNCATED AT 250 WORDS)