Logic is essential to the formation of legal texts, as the use of conjunction offers an insight into the whole logic of discourse (Smith and Frawley 1983, cited in Baker 2000: 192). This study investigates how the logic of legal documents and their translations are manifested by the use of conjunctions. Two legal documents are selected for comparison, namely prospectuses for the listed companies which belong to a business/financial-related legal genre, and the Hong Kong Companies Ordinance, a typical legislative genre. Based on the framework developed from Halliday and Hasan’s cohesion model, seven categories of cohesive conjunctives, namely additive, adversative, causal, temporal, continuative, hypothetical and clarifying conjunctives are investigated and compared through a corpus-based approach. Six types of translation methods, namely amplification, omission, conversion, diction, retention and negation are also identified and compared in the translation of each category of conjunctives. It is found out that 1) the conjunctive cohesion patterns of prospectuses are different from those of legislative texts in terms of both wording and frequencies, with the same case in their translations; 2) translation methods used in the translations of the two documents are different; 3) the translation of prospectuses tends to be implicit with the cohesive conjunctives, while the translation of legislative texts tends to be explicit. It is argued that possible reasons for the disparities include cross-linguistic differences and extra-linguistic factors such as generic, socio-cultural and translator differences. It is hoped that the findings will give translators some insights in applying cohesive conjunctives as well as different methods in their translations. The study also suggests that a corpus-based approach is applicable in examining cohesive conjunctives across two linguistic systems and different legal genres. The findings may serve as useful information for dictionary compilation and statistic-based machine translation program design.