There are several formulas for the quantitative determination of intrathecal IgG production: Reiber and Felgenhauer's formula (IgG(loc)), the Extended IgG index, Tourtellotte's formula (TOURT), Schuller and Sagar's formula (SCHULL), the IgG index, the Log IgG index, and Blennow and co-workers' formula (IGGPROD). To evaluate the utility of these formulas in the presence of blood-brain barrier (BBB) damage, we present the results from a study of serum and cerebrospinal fluid (CSF) samples from 125 healthy individuals, 18-88 years of age; 1072 consecutive patients without oligoclonal IgG bands (OCBs) in the CSF, 683 without BBB damage (CSF/S: albumin ratio < 9.8) and 389 with BBB damage (CSF/S albumin ratio 9.8-30); and 106 patients with definite multiple sclerosis (MS). The relation between the CSF/S albumin ratio and the CSF/S IgG ratio was remarkably linear in both healthy individuals (r = 0.95; P < 0.0001) and patients without oligoclonal bands in the CSF (r = 0.95; P < 0.0001). Therefore, IgG(loc) and the Extended IgG index, two formulas based on a nonlinear relation between the CSF/S albumin ratio and the CSF/S IgG ratio, yielded biased results (lower values) in the presence of BBB damage. TOURT and SCHULL also yielded biased (higher) values in the presence of BBB damage, probably because of incorrect constants in these formulas. There were no significant correlations between the CSF/S albumin ratio (i.e. the BBB function) and the IgG index or the Log IgG index, two dimensionless quotients for the detection of intrathecal IgG production, or between the CSF/S albumin ratio and IGGPROD, an empirical formula for the determination of intrathecal IgG production in mg/l.(ABSTRACT TRUNCATED AT 250 WORDS)