In my opinion, one of the things TSR did right during its existence was produce a wide variety of roleplaying games rather than endlessly churning out material for Dungeons & Dragons. There are several reasons why I think this is so, the most important of which is that these new games provided an outlet for players and game designers alike to try new things without feeling the need to inflict their desire for novelty on D&D. By this I meant that a great many of the slings and arrows D&D suffered over the years have been the result of people getting bored of its ways of doing things and then trying to change it to suit different tastes. Far better in my opinion is to play or to create new games specifically that speak to those tastes.
In my early years in the hobby, my friends and I regularly stopped playing Dungeons & Dragons when we felt the call of a different genre or even style. TSR happily provided us with plenty of other games from which to choose -- Gamma World, Gangbusters, Star Frontiers, Top Secret -- as did other game companies. We rarely stayed away from D&D for long, but I think our regular "sabbaticals" from it helped keep us interested in it over the long term. I still think that way today, which is why I see it as a good thing for a group to play many different RPGs (so long as it doesn't give way to the dreaded "gamer ADD").
I bring this up because one of the last non-D&D TSR roleplaying games I remember buying was 1985's Conan Role-Playing Game. Designed by David "Zeb" Cook, it's a complete game in a single box, consisting of a 32-page rulebook, a 16-page reference guide, and a 48-page guide to the Hyborian Age. I remain amazed at how often TSR produced RPGs whose rulebooks were 64 pages or under; it's frankly a thing of beauty -- all the moreso because, in most cases, these games didn't need to be any longer. Conan was a fairly "light" game, using a series of six "talent pools" to adjudicate every action a character takes in the game. The pools are broad in nature (Prowess, Fighting, Endurance, Knowledge, Perception, and Insight), under which are more specific talents that represent areas of unusual skill, like Brawling for Fighting or Reading/Writing for Knowledge. Talent checks are resolved using percentile dice and compared to a color-coded result chart, as was the style at the time. In cases where a character is attempting to do something an NPC is opposing (combat, for instance), two talents are compared, with the opposing talent subtracted from the active one to determine what line on the chart to read. Though the rulebook didn't do a particularly good job of explaining all this, in practice, the system was quite easy to use.
For my friends and I, though, what made Conan such a fascinating game, aside from its source material, was its character creation system, which encouraged players to think about who their character was and how he fit into the Hyborian Age. The "character folio" (i.e. record sheet) included a section entitled "story" with a series of blanks for the player to fill in, like Mad Libs. So, it would say "(Character Name), (Sex) of (Father) and (Mother), was born in the land of (Homeland). (Character Name) grew (Appearance). As a youth, (Character Name) learned (Talents)." And so on. The idea is to frame one's character abilities into something that is coherent, interesting, and fitting for an adventurer in a Robert E. Howard yarn. It's a simple thing really, but I can't stress how revelatory this approach was to my friends and I. We found it so much easier to get into the spirit of things this way.
Of course, it probably helped that Conan Role-Playing Game was clearly a labor of love on the part of David Cook. His fondness for the stories of Conan is well known and this game is like a love letter to the tales of the Cimmerian. For example, the game's magic system is mechanically loose and difficult for characters to master, with many opportunities for both disaster and long-term consequences. This is, of course, as it should be, for magic in the Hyborian Age is usually a dark affair not to be trifled with. This didn't stop one of my friends from creating a sorcerer character, named Talon after the twisted bird-like claw he had on one hand -- a reminder of his playing with things Man Was Not Meant to Know. Just as well presented is the game's setting, which is presented in the form of a faux notebook written by an archeologist named Ervin H. Roberts who was convinced that there was "an age undreamed of" in prehistory. It's a little silly, of course, but in a good way that I think encourages fun rather than hinders it.
I vaguely recall that TSR produced a handful of modules of to support Conan Role-Playing Game (in addition to those produced for AD&D), but I may be misremembering. Regardless, I don't believe the game did very well for the company, or at least it didn't do well enough, because it wasn't on store shelves for long. I think that's a shame, because Conan was well-presented, straightforward, and fun. It could never replace Dungeons & Dragons in the eyes of my friends and I, but that's hardly the benchmark by which a game ought to be judged. Unfortunately, I think that's exactly the benchmark TSR used, which is why most of their non-D&D RPGs had extremely short lives. A pity.