Objectives: Bar attachments have been established in implant dentistry. Contrary to prefabricated and individual fabricated metal or electroplated matrices semi-precise resin matrices are economical and easy replaceable. Manufacturers offer various retention forces [RF]. The aim of this in vitro study was to analyze and compare retention characteristics of different prefabricated resin matrices in implant dentistry. Methods: 4 bar systems (Dolder-(U-shaped-)bar-attachments [DBA; C&M], parallel-bar-attachment [PBA], snap-bar-attachment [SBA] and resilient-snap-bar-attachment [RSBA; Bredent]) were tested for retention of the corresponding specific resin matrices. For each system patrices were casted using prefabricated acrylic components (Heraenium, Heraeus Kulzer) and torqued to an implant (RN, Straumann). For each system 3 coloured matrices (n=3x10) with different RF (A, B, C) were tested. RF of 3 systems are indicated by the manufacturers (4/6/8N). DBA quotes 'light'/'middle'/'heavy'. In an universal-testing-machine matrices were removed 10 times from the patrices in (v=50mm/min, axial). Force-deflection-graphs were made. If possible mean RF were compared to manufacturer's data and statistically analyzed (p0,05). Results: RF differed within the groups: DBA 3,6±1,8N to 17,6±1,0N, PBA 5,6±0,6N to 7,9±0,8N, SBA 4,2±1,2N to 7,5±1,0N, RSBA 5,8±1,8N to 8,9±1,6N. Comparing manufacturer's data partially significant differences were recorded: PBA 93,3-152,5%, SBA 70-122,5%;RSBA 108,3-145%. Manufacturer's data of RF were different: PBA-matrixA (4N) 6,1N, PBA-matrixB (6N) 5,6N; SBA-matrixA (4N) 4,9N; RSBA-matrixB (6N) 4,2N. Conclusion: RF of the matrices of most systems showed differences to manufacturer's data. Purposeful application of some systems according to RF seems to be doubtful and unpractical. Schlagwörter: implant dentistry, attachments, bar, retention