First off, let me be clear. This is a very good book. It does a great job of describing the relational model: what it is, how it works, and where SQL converges/diverges from it.
So why only 3 stars? Because of the incessant sniping at SQL the whole way through. Date has been vocal for years about SQL falling short of the relational model, and exhorting the DBMS world to step up. He (and Hugh Darwen) also provided a comprehensive specification for a next generation DBMS that fully realised the power of the relational model. For whatever reason, the market hasn't delivered the vision in product.
Whether because of that, Date seems to have become obsessed with pointing out SQL shortcomings against the relational model. And that is a real shame. This book would have been /so/ much better had Date focused on real-world problems, shown how the relational model elegantly handled them, and then providing solid advice on how best to solve the same problem using SQL, shortcomings and all.
Doing so would have made the book even more useful and removed the frustrated undertone. Paradoxically, it would likely also have served Date's goal to expose SQL's shortcomings and perhaps motivate improvement from the vendors.
So ultimately a good book that could have been great. It nevertheless remains an excellent description of the relational model, far more accessible than either "Foundations of Database Systems" (as mentioned elsewhere) or "The Third Manifesto".