waittime function precision

Hi,
I've used waittime function for pretty long time in cases, where I needed to have a delay in my program.
But recently a question appeared - how precise it is and are there any alternatives that allow to make delays for less than 1/100 of a second.
For an experiment I wrote a short program:

int main (int argc, char *argv[])
{
_MI_Time WaitTime ;
struct timeval tp1, tp2 ;
struct timezone tpz ;
decimal (9,3) Duration ;
decimal (19,6) StartTime, EndTime ;
int i ;

for (i = 1; i < 10; i++)
{
gettimeofday(&tp1, &tpz) ;
StartTime = tp1.tv_usec ;
StartTime = tp1.tv_sec + (StartTime / 1000000D) ;

mitime(&WaitTime, 0, 0, 0, i) ; /*hours,minutes,seconds,1/100*/
waittime(&WaitTime, _WAIT_NORMAL) ;

gettimeofday(&tp2, &tpz) ;
EndTime = tp2.tv_usec ;
EndTime = tp2.tv_sec + (EndTime / 1000000D) ;
Duration = (decimal (9,3)) ((EndTime - StartTime) * 1000D) ;
printf("Delay = %d ms, waittime: %D(9,3) ms\n", i * 10, Duration) ;
}
return(0);
}

So it makes 10 cycles producing delays (theoretically) 10, 20, 30... 100 ms.

And here's the result:
Delay = 10 ms, waittime: 18.480 ms
Delay = 20 ms, waittime: 38.960 ms
Delay = 30 ms, waittime: 59.440 ms
Delay = 40 ms, waittime: 69.984 ms
Delay = 50 ms, waittime: 50.552 ms
Delay = 60 ms, waittime: 84.936 ms
Delay = 70 ms, waittime: 83.336 ms
Delay = 80 ms, waittime: 109.912 ms
Delay = 90 ms, waittime: 119.136 ms

Isn't it strange?
I have no reasons to think that gettimeofday gives invalid resuts and the calculations are pretty simple...
Is it a fundamental feature of waittime?

Jevgeni