Well, Phoenix, I have three relevant works, being the aforementioned A&G and Gildersleeve, and New Latin Syntax by E. C. Woodcock as well(indeed, an odd name, at least nowadays).
As to the merits of the books, one is to note that A&G and Gildersleeve are of the same nature: reference grammars. Therefore, they extensively treat on all
relevant aspects of the language, yet are scanty as to explaining in depth topics which might be considered difficult. For this reason, these books appear accommodated to a more "advanced" audience, in the number of which audience surely you are to be counted.
The question is then: which book, provided I don't want to purchase both, comes most recommended? In short, I'd opt Gildersleeve. Notwithstanding A&G's excellence, as it treats -much in the same fashion as Gildersleeve by the way- all facets of syntaxis, Gildersleeve has a more systematic approach. On average, A&G seems to make the impression of a comprehensive collection of all rules regarding Latin syntaxis, somewhat irrespective of the underlying system of syntaxis, that is, somewhat dissociated from "the big picture", whereas Gildersleeve -though not in itself more expansive- presents the information in a more sensible, that is with more respect towards the system of syntaxis, order. In this sense, Gildersleeve has an added value over A&G.
As for Woodcock, this book -far from a reference work- seeks to elucidate in detail more specific subjects of syntaxis, discussing them in depth. Therefore, it doesn't treat all
conceiveable topics, yet selects the most important ones, and treats them thoroughly. Thus, one might easily choose
between A&G and Gildersleeve, yet regardless which one one chooses, Woodcock is, in my opinion, a very worthy addition, on account of its distinct nature.