In 21 patients with advanced periodontitis, 39 teeth exhibiting class II (n = 21) and class III (n = 18) furcations were treated by the guided tissue regeneration technique using expanded polytetrafluoroethylene (ePTFE) membranes (n = 20) or bioabsorbable barriers (n = 19), respectively. Clinical parameters were assessed before and 6 months after surgery. Presurgically and 6 months postsurgically, 35 pairs of standardized bitewing radiographs were taken. Using subtraction radiography, gain of bone density within furcation areas was assessed. Eighteen radiographs showed sufficiently accurate geometry to be analyzed by subtraction. Within the defects suitable for subtraction, the average gain of vertical attachment assessed was 1.35 +/- 1.27 mm in the class II furcation group and 1.58 +/- 1.37 mm in the class III furcation group. The average horizontal attachment gain in the class II furcation group was 1.96 +/- 0.59 mm. No statistically significant differences were observed between results after GTR therapy using non-resorbable and bioabsorbable barriers. Radiographic bone gain as assessed by subtraction analysis correlated with vertical (r = 0.458, P < 0.025) and horizontal (r = 0.734, P < 0.005) attachment gain. A statistically significant number of more radiographs for maxillary molars were not suitable for subtraction analysis than mandibular molars (P < 0.05). Further, statistically more radiographs that were taken with potentially unstable support of the filmholder were not suitable for subtraction analysis than those with stable support (P < 0.05). There is a statistically significant correlation between clinical improvements and bony fill within furcation defects. Only 18 of 35 pairs of radiographs were suitable for subtraction analysis. Subtraction analysis of maxillary molars seems to be more difficult than assessment of radiographic bone changes in mandibular molars. Potentially stable support of the filmholder seems to be a condition to provide radiographs suitable for subtraction analysis.