2.1. It's All in Osterbrock
The logical starting point for most of us is the wonderful text by Osterbrock (1989; AGN2). It really is true that "it's all in Osterbrock," but that does not necessarily mean that this text provides a simple, straightforward introduction to the subject of nebular astrophysics. In fact, it can serve in that role, but I think a few words of guidance will help a great deal.
From my experience in teaching from AGN2, the greatest difficulty arising from first encounters with this text lies in one of the book's great strengths. AGN2 is organized around physical principles, and is extremely thorough in its treatment. While this allows one to find material easily and to be confident that one really only needs one book, this makes it difficult for the first-time reader. For example, after the discussion of the calculation of spectra (chapter 4) and before the discussion of the comparison of theory with observations (chapter 5), are several sections with relatively detailed descriptions of specific problems (e.g., the Bowen resonance-florescence mechanism for O III).
The very simple solution to this problem is to be aware of the structure of the text, and to be confident that is it possible to skip over many of the chapter sections without fear that the following material will be incomprehensible. Perhaps this is obvious to most, but I have found this simple advice to be invaluable for many AGN2 neophytes.
A second simple suggestion that I have found very effective is to remind students that an HII region is an extreme example of a non-LTE problem. Since many HII region observables are the result of processes associated with the electron gas which has a thermalized velocity distribution, students are tempted to apply their LTE instincts when approaching HII region problems. Usually this has disastrous results (e.g., naive models of gas temperatures dropping as distance2 from the exciting star). It is important to remember that the energy input source is the absorption of radiation from the exciting star, and to separate highly non-LTE phenomena (e.g., the recombination emission spectrum of H and He) from LTE phenomena (the collisional excitation of the lowest levels of the heavier ions).