Nowadays, translation may be the bottleneck of the pretended information globalisation. While surfing the Internet, for instance, sometimes we come across languages and characters we don't understand. Statistical machine translation (SMT) constitutes a research sub-area of machine translation (MT) that has recently gained much popularity. In fact, this technology has experienced real growth motivated by the development of computer resources needed to implement translation algorithms based on statistical methods. This PhD thesis focuses on the SMT framework and primarly on the definition and experimentation of novel algorithms for building a correct structural reordering for translated words. Moreover, challenging techniques regarding language modeling and system combination are successfully applied to state-of-the-art SMT systems. This dissertation should shed some light on the SMT approach and on the word ordering challenges and should be specially useful to natural language processing researchers having non or some expertise in machine translation.