First, how are your EMG voltages being measured? Are they peak-to-peak values, integrated EMG, or root-mean-square? Regardless, the values will be affected by the impedance of the electrode:skin interface and that of the tissues between the skin and the underlying muscle fibers and these will always vary to some degree from person to person. The lower the impedances, the lower the resting EMG amplitude. This assumes that underlying muscle fibers are electrically silent at rest. What you are really measuring with your "resting voltages" is noise of the system. Also, EMG amplifiers employing active electrodes will generally yield lower resting EMG (i.e., lower noise) than that of systems using passive electrodes.

It is hard for me to tell what is normal for your EMG system. Unless you are using active electrodes, I recommend that you invest in an impedance meter to determine the impedance between pairs of electrodes. Shoot for an impedance of ~10 kOhm or less. If you are over this number, trying cleaning (and maybe gently abrading) the skin again. And make sure the electrodes are securely attached to the skin.

Final comment. It sounds like you may be wanting to normalize your shivering EMG activity to your so-called resting voltage. I would instead normalize to the amplitude of the EMG activity measured during a maximal isometric voluntary contraction.

Thank you very much for your answer. We are using a PowerLab system and obtaining the voltages from the RMS.  We are also standardising to an MVC for each muscle, as you suggested, and subtracting the resting value from the actual value witnessed during shivering, so again you are correct in that we are simply removing the noise. I don't have a background in EMG so it concerned me that there was a large difference between subjects.  

The set up is virtually identical to the attached paper, and I have also attached an image which shows 1) the voltages and some of the waves I obtained, and 2) how the voltages will be used to express shivering intensity as a %MVC.

Thanks again for your answer.   

It is not uncommon to have a decent amount of variation among subjects in baseline EMG amplitude. Ensuring that electrode impedance is ~10 kOhm or less for all subjects will help reduce it.