The Basic Averaged DC-RMS VI uses a weighted average to compute the RMS value of the input signal. The weighting used is a window shape, and serves to emphasize the center portion of the signal over the endpoints. For this type of method to accurately identify the RMS value you should input several cycles of the tone. Have you tried using the Averaged DC-RMS VI instead? This VI will allow you to specify an averaging time that is different than the length of the input signal. This way you can feed data to the VI in a loop, and the VI will accumulate the data until the averaging time has been exceeded, at which point the VI will produce an output value.
I suggest looking at the following example to play around with these parameters so you can get a feel for how they work.
...\LabVIEW\examples\measure\maxmpl.llb\Advanced DC-RMS Measurement.vi
In the "Test Signal" section change the square amplitude and noise level to be 0. That will leave a 50Hz sine tone as the signal being measured. Also change the "output function" to be "RMS Only" and the "window for RMS" to be "low side lobe". Notice that the RMS value varies between 2.3 and 2.7 volts. If you change the "averaging time" to be 100ms (from the default 10 ms) notice how much more consistent the RMS value is. The longer you average the better the measurement will be. Basic Averaged DC-RMS uses the length of the input signal as the averaging time.
-Jim