We developed an automatic piano performance system calledPolyhymnia that is able to generate expressive polyphonicpiano performances with music scores so that it can be usedas a computer-based tool for an expressive performance.The system automatically renders expressive piano musicby means of automatic musical symbol interpretation andstatistical models of structure-expression relations regarding polyphonic features of piano performance. Experimental results indicate that the generated performances of various piano pieces with diverse trained models had polyphonicexpression and sounded expressively. In addition, the models trained with different performance styles reflected thestyles observed in the training performances, and they werewell distinguishable by human listeners. Polyhymnia wonthe first prize in the autonomous section of the PerformanceRendering Contest for Computer Systems (Rencon) 2010.