Surveying the complaints people have about BasiliskII and SheepShaver I think what's really going on with them falls into two categories:
1: They're too complicated, and
2: Bad luck with OS-specific ports or builds.
In reference to #1, SheepShaver and BasiliskII (BasiliskII in particular) are flexible pieces of software. They work with numerous ROM images, they have oodles of run-time configurable options, and they support *much more* functionality then vMac ever did. (Serial port emulation, SCSI manager, direct drive access...) Most of the development work for these emulators goes toward improving the functionality or performance of them, not towards making them simple to use. And the fact that it's illegal to "bundle" them with known-reliable-and-tested ROM and operating system images means that no matter how good a GUI you slap on top they're always going to have something of a steep learning curve, and will work better for some people then others.
vMac, by contrast, has nearly every option set at compile time. A given binary emulates a single machine, supporting a single ROM image, and emulating a very limited set of peripherals. It's nowhere near as ambitious, and thus it's worlds simpler. It's the difference between flying an F-15 Eagle and a Piper Cub. The fact that the Cub is easier to fly doesn't necessarily make it a better airplane.
With reference to point #2, I started using BasilikII *way back* in 1999 on Linux. (The original non-JIT processor emulation managed somewhere in the ballpark of LC II-level performance on the 133Mhz Cyrix machine I ran it on. Those were the days.) It was hard to compile (this problem got much worse in the months after the JIT compiler debuted), the methods for dealing with HFS images were primitive at best, if you wanted to use networking you had to go through some positively baroque processes manually... basically, all the sort of things that'd make your average Mac user's head implode. Lately, however, the Linux versions have gotten *very good*. I've had basically zero problems since they ported BasiliskII to use SDL for video output (there were some long-running annoyances prior to that), and Sheepshaver is likewise trouble-free.
The software was originally focused on UNIX/Linux platforms with X11. The Windows version was always something of an afterthought, and although it seems like OS X is where the most *user* interest lies now the quality of the code going into the OS X conversions just don't seem that great. (I get the feeling that some of the most competent core developers either don't care about non-Linux platforms or have moved on.) It's a pity, but I'd still say it's a bit unfair to say that SheepShaver/BasiliskII "suck". It's more accurate to say that the quality control of the released binaries for those OS'es ports suck. :^b
(You do have to consider that when you download a binary build off a forum, which seems to be the way to get Sheepshaver these days, you're looking at something someone compiled out of CVS, fired up for long enough to verify that it doesn't choke *immediately*, and then unleashed on the world with hopes that the last dozen fixes checked in for obscure problems actually worked. To get tested bug-free software you have to pay beta-testers, which means you have to pay for the software. With free open-source projects like this it's not really free: You *are* paying for it with your beta-testing time.)
If vMac does everything you want it to do use it. It is *simpler*. I just have my doubts that if vMac were to actually grow to cover even a subset of the feature territory that BasiliskII does that its quality will be any better.
It would be interesting if vMac were to try harder to be a full cycle-accurate "all-hardware" emulator with its Mac II support, with the ultimate goal being to emulate an MMU-equipped MacII-family machine well enough to run A/UX. (edit, not OS X, duh) But honestly, I'd almost bet on MESS beating them to it.