Purpose: The purpose of this study was to systematically review clinical studies on the incidence of abutment screw loosening in single-implant restorations with different implant-abutment connection geometries. Materials and Methods: The literature search was conducted using several electronic databases. Specific terms were used for the database search, which spanned the years 1990 to 2006. The search was augmented by using the option of "related articles" as well as hand searching of references and relevant journals. Relevant studies were selected according to predetermined inclusion and exclusion criteria. Agreement between reviewers was determined by using Cohen's kappa coefficients. Three-year complication-free rates (survival proportions) were calculated with the aid of a survival function, assuming constant failure rates. Summary estimates per group for complication-free rate after 3 years (M-estimator) were calculated using Tukey's biweight estimator. Results: The initial database search yielded 1,526 relevant titles. After the subsequent filtering process, 27 studies were finally selected. Interexaminer agreement ranged from good to perfect. The external-connection group comprised 12 studies following 586 single-implant restorations for a mean follow-up time that ranged from 3 to 5 years. The estimated percent of complication-free single-implant restorations after 3 years was 97.3% (95% CI: 95.6-98.3). The internal connection group comprised 15 studies following 1,113 single-implant restorations for a mean follow-up time that ranged from 3 to 10 years. The estimated percentage of complication-free single-implant restorations after 3 years was 97.6% (95% CI: 96.5-98.3). Conclusion: The results show that abutment screw loosening is a rare event in single-implant restorations regardless of the geometry of implant-abutment connection, provided that proper antirotational features and torque are employed. Palabras clave: implant complications, implant geometry, meta-analysis, screw loosening, single implants