Purpose: The aim of this randomised controlled double-blinded clinical trial was to determine the efficacy of a new cross-linked membrane (VN) in guided bone regeneration (GBR) around exposed dental implants compared to a native collagen membrane (BG). Material and methods: A total of 16 patients in need of implant treatment at two different sites with osseous defects were planned for this split-mouth study. After inserting the dental implants, periimplant defects were treated according to the GBR technique using a VN membrane with prolonged resorption time in the randomised test site and a BG membrane in the control site. After a healing time of 6 months, mucoperiosteal flaps were elevated for the evaluation of the primary (vertical bone fill [ΔDL] and quality of newly formed tissue [QT]) and secondary outcome variables (infrabony defect height [DH], defect width [DW], defect depth [DD] and augmentation depth [AD]) and the sampling of biopsies apical to the implant shoulder. Results: A total of 16 patients fulfilled the initial non-surgical inclusion and exclusion criteria. However, the study was discontinued early after 9 surgically treated patients because unacceptable safety issues arose and severe infection related to the VN membranes. The VN membrane revealed statistically significantly more soft tissue dehiscence than the BG membrane (56% and 11%, respectively, P = 0.0455). In 3 of these 9 patients the VN membrane had to be removed due to infection early after the first follow-up visit. For the statistical analyses these sites were designated as the value of the baseline. The mean ΔDL values were 1.8 ± 1.6 mm at the VN site and 4.7 ± 3.3 mm at the BG site. The ΔDD values were 0.6 ± 1.0 mm and 1.1 ± 1.2 mm, respectively, and reached statistical significance (P = 0.0208, CI 95% = -2.9 [-5.2;-0.6]). The corresponding linear defect fill (DF) values were 44% and 78%, respectively. The clinical assessment of QT showed comparable median values at sites treated with VN (3, interquartile range: 0; 3.5) and BG (3, interquartile range: 3; 4) without statistical significance. The histomorphometric analysis showed an average area density of 24.4% (SD 10.3, range 8-35%) newly formed bone at the test sites and of 35.0% (SD 20.6, range 8-60%) at the control sites. The histological data showed only some trends and did not reach statistical significance. Conclusion: In the present study, the VN membranes with prolonged resorption time demonstrated significantly more adverse events and insufficient bone regeneration compared to the native BG membranes and no advantages in favour of the VN membranes were detectable. Schlagwörter: clinical research, collagen membrane, cross-linking, guided bone regeneration