Preload loss can favor the occurrence of implant-abutment interface misfit, and bacterial colonization at this interface may lead to implant failure. The aim of this study was to evaluate the preload loss and bacterial penetration through the implant-abutment interface of conical and external hexagon connection systems subjected to thermal cycling and mechanical fatigue (TM). Four different implant-abutment connection systems were evaluated (n=6): external hexagon with universal post, Morse taper with universal post, Morse taper with universal post through bolt, and locking taper with standard abutment. The assemblies (implant-abutment) were subjected to a thermal cycling regimen (1,000 cycles of 5°C and 55°C) and to mechanical fatigue (1.0 million cycles, 1.0 Hz, 120 N). The assemblies were immersed in Tryptic Soy + Yeast Extract broth containing Streptococcus sanguinis and incubated at 37°C and 10% CO 2 for 72 h. Detorque values were recorded. The bacterial penetration was assessed and the abutments were observed by scanning electron microscopy. The preload data were analyzed statistically by two-way ANOVA and Tukey's test at 5% significance level. All screw abutment systems showed significantly higher (p<0.05) detorque values when subjected to TM and all conical systems presented bacterial penetration. The results show no relationship between the preload loss and the bacterial penetration.