I reviewed the first edition of this book in Knowledge Engineering Review 20(4), and thought it was a very good book on which to base a one semester introduction to machine learning. This edition is even better. The new edition is more than 150 pages longer, and the additional pages reflect a considerably wider scope. Rather than support vector machines being relegated to a section of a chapter on discrimination, there is now a whole chapter on kernal machines, better reflecting the interest in the field. Similarly probabilistic graphical models now have their own chapter. There is a new chapter on Bayesian estimation, and the chapter on experimental technique, which includes very valuable material on assessing the outcome of experiments, is much expanded. Overall, these are important upgrades to the book, and I think it is significantly improved. I am not sure that the new edition is any more likely to get me to teach the one semester introduction to machine learning that the first edition made me want to teach, but if it does, then Introduction to Machine Learning will definitely be the textbook.