Solid-state NMR spectra have historically been assigned using simple relationships between NMR parameters, e.g., the isotropic chemical shift, and aspects of the local structure of the material in question, e.g., bond angles or lengths. Density functional theory (DFT) calculations have effectively superseded these relationships in many cases, owing to the accuracy of the NMR parameters typically able to be calculated. However, the computational time required for DFT calculations may still be prohibitive, particularly for very large systems, where structure-spectrum relationships must still be used to interpret the NMR spectra. Here we show that, for calcined aluminophosphates (AlPOs), structurespectrum relationships relying on either the mean P-O-Al angle or the mean P-O distance, both suggested in previous literature, provide a poor prediction of the 31 P isotropic shielding, iso, calculated by DFT. However, a relationship dependent on both parameters yields predicted iso in excellent agreement with DFT, with a mean error of ~1.6 ppm. The predictive ability of the relationship is not improved by introducing further parameters (many used in previous work) describing the local structure, suggesting that the twoparameter relationship is close to an optimum balance between accuracy and overparameterisation. The ability to predict accurately the outcome of DFT-level calculations will be of particular interest in cases where the actual calculations would be impractical or even impossible with current computational hardware, or where many such calculations are required quickly.