Tells are the one part of poker theory that haven't yet seen a spectacular increase in knowledge since the poker boom. This is probably because most poker over the last decade has been played over the internet behind a computer screen.
This could well be the book we've all been waiting for.
A big problem with the early work on tells, Mike Caro's in particular, is that it really isn't a proper science at all. Caro categorises tells from "actors" -- for whom strong means weak and vice versa -- and "non-actors" who tell you about the real strength of their hand. The trouble is that such a broad theory is in effect irrefutable, since any combination of mannerisms and end-resulting hand strength can be explained. As philosopher of science Karl Popper points out, a theory that can explain any observation without potential refutation -- e.g. Marxism, Freudian psychoanalysis -- is at best pseudo-scientific nonsense.
Elwood's book is much better. For a start he stresses the overwhelming importance of the situation (an opinion much confirmed by the field of social psychology), the need for developing a statistically robust level of observational data over time, and bases his theories on the psychology of unconscious human behaviour. At all times he stresses that poker tells are just another tool in a skilled player's arsenal; not a strategy for basing your entire game around.
But at this price the book is a clear winning investment for any frequent live player or even an online specialist entering a casino environment for the very first time.