BACKGROUND The purpose of this study was to evaluate the influence of root trunk concavities on guided tissue regeneration in humans and to determine if a modification in membrane design could improve the results. METHODS Ten patients who provided 10 pairs of similar Class II furcation defects were recruited for the study. Each defect was randomly assigned to treatment with either a normal expanded polytetrafluoroethylene (ePTFE) membrane (NM) or a modified ePTFE membrane (MM). The MM were made by removing the collar from an NM, cutting it into 2 mm segments, and suturing each of the segments to the collar of each MM. Following basic therapy, baseline clinical measurements were recorded using an acrylic stent, including probing depth (PD) and clinical attachment level (CAL). Full flaps were elevated, the defects were carefully debrided, and the root surfaces were decontaminated with manual and ultrasonic instruments. Hard tissue measurements were performed during surgery to determine vertical (VD) and horizontal defect (HD) depths. Then, NM and MM were sutured to their corresponding tooth and the flaps sutured. Membranes remained in position for 6 weeks. After 1 year, all sites were re-entered and soft and hard tissue measurements were recorded. RESULTS After 1 year, there were no statistically significant differences (Mann-Whitney test) between NM and MM groups in percentages of PD reduction (P= 0.27), CAL change (P= 0.26), or VD resolution (P = 0.13). However, percentage of HD resolution (NM: 24.3+/-11.8; MM: 44.8+/-16.0) was significantly different between groups (P= 0.008). CONCLUSIONS The modified membranes resulted in greater horizontal resolution of Class II furcation defects. The collars of the membranes should be modified to improve results when root trunk concavities are present.