Purpose: The aim of this study was to evaluate, with a new gas flow technique, leakage at the implant/abutment junction in systems with four different connections. Materials and Methods: Five Brånemark System, five One Morse, five Intra-lock System, and five Ankylos Plus implants and abutments were used. A hole was drilled in the apex of each implant to allow gas to flow through the connection from negative to atmospheric pressure. The gas flow was calculated (slope of pressure decrease, in hPa.s-1). Each connection was tested after both manual and key tightening. Statistical analysis was performed on a generalized linear model with repeated measurements. The significance level was set at α = .05. Results: A global significant difference was observed between the various systems (P = .0001). After manual tightening, gas leakage was (Ln[hPa.s-1], means ± standard deviations): One Morse: 0.20 (± 1.70); Brånemark System: -4.56 (± 2.61); Intra-lock: -4.31 (± 4.17); Ankylos Plus; -7.59 (± 0.76). After key tightening, mean values were: One Morse: -2.51 (± 2.72); Brånemark System: -7.23 (± 1.01); Intra-lock: -7.76 (± 0.50); Ankylos Plus; -7.73 (± 0.62). Conclusion: This study confirms that gas flow is an appropriate method to assess connection leakage. Ankylos Plus connection leakage was very low when the assembly was tightened manually. Among conical connection systems, low (Ankylos Plus) and high (One Morse) leakage was observed. This gas flow study suggests, therefore, that connection design is not the most important parameter for implant/abutment connection leakage. Palabras clave: abutment, connection, dental implant, gas flow, implant-abutment interface, leakage