The aim of this study was to investigate the reliability of the assessment of clinical furcation parameters (horizontal attachment levels [CAL-H], class of furcation invasion). Replicate measurements of CAL-H and furcation class were performed within 14 days in 420 molars of 105 patients with advanced periodontitis using a Nabers- (n = 50), a TPS- (n = 30) and a PCPUNC15-probe (n = 25). Validity of clinical assessments was assessed by intrasurgical measurements. The standard deviation of single measurements (s) was calculated as a measure of the reproducibility of CAL-H measurements and weighted kappa-coefficients (κw) to estimate the agreement of furcation class assessments. The s ranged from 0.55 to 1.13 mm (Nabers), 0.55 to 1.02 mm (TPS), and 0.58 to 1.11 mm (PCPUNC15). For all probes, a statistically significantly smaller measurement error was observed in buccal and lingual sites than in mesiolingual and distolingual furcations (P < 0.005). The κw ranged from 0.59 to 0.89 (Nabers), 0.50 to 0.80 (TPS), and 0.53 to 0.72 (PCPUNC15). Multiple linear regression analysis identified distolingual location, probing depth (PD) and CAL- H as factors influencing the variability of CAL-H measurements. Whereas there was no statistically significant difference between pre- and intrasurgical CAL-H measurements using the Nabers probe, the TPS and PCPUNC15 probe underestimated CAL-H for distolingual furcations (P < 0.025). Using the Nabers probe, no asymmetries between pre- and intrasurgically obtained class of furcation involvement were revealed, while the TPS and PCPUNC15 probe underestimated furcation degrees (P < 0.1). Multiple linear regression analysis identified distolingual location and height of furcation, as well as PD, vertical attachment level (CAL-V) and type of probe, as factors influencing the validity of CAL-H measurements. Clinical diagnosis of furcation lesions using the 3 mm incrementally marked Nabers probe provides reproducible and valid information about furcation invasion. J Periodontol 1998;69:328–336.