OBJECTIVE To investigate the sensitivity level of resonance frequency analysis (RFA) in the identification of contact loss at compromised bone-implant interfaces. METHODS Twelve cylindrical acrylic experimental models housing a perpendicularly positioned dental implant were generated. Two different compromised bone-implant interfaces to simulate the loss of bone-to-implant contact were created in 8 models, whereas the resting 4 simulated the intact relationship between implant and bone. Each implant received RFA and subjected to serial sectioning to evaluate bone-acrylic interface microscopically. Implant stability quotient values were statistically compared among groups. Correlation level between the implant stability quotient values and bone-acrylic interface conditions was assessed. RESULTS RFA was not sensitive to recognize the lack of horizontal contact (approximately 0.276 mm), but it was able be to discriminate the additional loss of approximately 0.1 mm at the bone-implant interface. CONCLUSIONS RFA is not a reliable method in the assessment of the initial contact loss at bone-implant interface.