31P NMR spectroscopy provides a means of monitoring intracellular pH (pHi). In order to determine the significance of observed pH changes, an assessment of the measurement uncertainty is required. This paper considers the possible sources of error involved, firstly in chemical shift assignment and, secondly, in relating chemical shift to pH. A statistical analysis of these errors is presented and an expression for the accuracy of pH measurements obtained using in vivo 31P NMR spectroscopy is derived. Computer simulations and clinical measurements are considered to determine the likely error in clinical measurements. A comparison of pH measurement methodology using a centre-of-gravity calculation, Gaussian and Lorentzian fitting is also presented and discussed.