Batch processes usually consist of multiple phases with different characteristics. Many research works have been done, focusing on process monitoring as well as quality prediction using phase information. However, few of them consider transitions from phase to phase, although transitions widely exit in batch processes and have non-ignorable impacts on product qualities. In the present work, a phase-based PLS method utilizing transition information is proposed to give both online and offline quality predictions, based on the understanding of the way transitions impact product qualities. First, batch processes are divided into several phases. Regression parameters other than prior process knowledge are utilized in phase division. Then both steady phases and transitions which have great influences on qualities are identified as critical-to-quality phases using statistical methods. Finally, based on the analysis of different characteristics of transitions and steady phases, an accumulative algorithm is developed for quality prediction. The application to an injection molding process shows the effectiveness of the proposed algorithm in comparison with the traditional MPLS method and the stage-based PLS method.