Latency is always a problem - depending on how you record. If you don't monitor through Cubase, then latency doesn't matter. If you DO monitor through Cubase, you'll be at the mercy of whatever latency your system can work with. Similarly, if you want to play VSTi's in real time, then you'll also be at the mercy of the same latency, but only inasmuch as what you hear played back will be delayed by the latency value. What is actually being recorded (ie. midi data) will line up perfectly to your click.
With audio, the same general idea applies. If you have a latency of, say, 100ms, then everything you hear back while monitoring through Cubase will be 100ms late. However, the audio signal goes INTO Cubase at the right time, so it should be in time. Latency is essentially a figure that describes how long it takes to process the in-coming signal and spit it back out again. You should not have to move any of your wavs so that they are in their proper time, unless they were not played as such in the first place - no matter how high your latency is. If you do, there is another problem somehwere.
As PsyCoNo said, latency is determined by a number of factors. To a small degree, the overall speed and efficiency of your system will affect it. The two biggest factors are the soundcard (and the accompanying drivers, of course...) you use, and where your buffer size is set to in Cubase. With my lowly Celeron 466 computer, and my good quality Delta44 soundcard, I can get as low as 8ms of latency, which is perfectly fine for monitoring through Cubase, or monitoring real time playing of VSTi's. This isn't a fully reliable figure, though, as adding more tracks into a project will require an adjustment to my buffer size, which means that I have to settle for higher latency. Once I've recorded about 20 tracks, I'm pretty much stuck with about 60ms of latency.
Chris