The aim of the study was to compare the in vivo early bacterial plaque colonization of 3 different guided tissue regeneration (GTR) membrane materials using a morphological (scanning electron microscope) method. Rectangular-shaped strips were cut from 3 periodontal membranes (expanded polytetrafluoroethylene, polyglactin 910, and polylactic acid) and glued to the buccal aspect of removable acrylic devices, which were applied to the molar-premolar region of the upper quadrants in 8 dental students. Each device held 3 strips: one ePTFE, one polyglactin 910, and one polylactic acid. The surface roughness of each membrane material was measured by means of a laser profilometer. During a 24-hour period, the students had to refrain from any oral hygiene procedures and did not use chlorhexidine mouthrinses. In each subject, one device was removed after 4 hours and the other after 24 hours. After removal, the devices were placed in a 2.5% gluteraldehyde solution to fix the membranes, which were then processed for SEM analysis. Fifty-four microscopic fields (at 200× magnification) were randomly selected and analyzed in each strip. Magnification was increased to determine the presence of bacterial morphotypes. The presence or absence of bacteria was assessed in a binomial fashion. In such a system, the field was bacteriapositive when bacteria constituted the deposits covering the surface of the membrane. The microscopic field was considered bacteria-negative when no bacteria were present. Bacteria-positive fields showing rods and filaments as prevalent bacterial morphotypes were recorded as rod-positive fields. A different pattern of plaque accumulation was demonstrated on different membrane materials. The 4-hour results indicated a statistically significant difference (P = 0.008, ANOVA) in the proportion of bacteria-positive fields among the 3 membranes; a greater amount of bacteria was demonstrated on the ePTFE membrane compared to the other 2 membranes. At 24 hours, the difference in the proportion of bacteria-positive fields was statistically significant (P = 0.002, ANOVA); a lesser amount of bacterial plaque was present on the polylactic acid membrane compared to the ePTFE and polyglactin 910 membranes. No difference in the proportion of rod/bacteria-positive fields was demonstrated among the 3 membranes at either 4 or 24 hours. It was concluded that quantitative differences in early plaque accumulation on various membranes seem to be related to the textural and structural characteristics of the surface, which is not adequately represented by the surface Ra value measured with a profilometric instrument. J Periodontol 1998;69: 1193–1202.