We test the efficacy of the energy-balance spectral energy distribution (SED) fitting code MAGPHYS for recovering the spatially-resolved properties of a simulated isolated disc galaxy, for which it was not designed. We perform 226,950 MAGPHYS SED fits to regions between 0.2 kpc and 25 kpc in size across the galaxy's disc, viewed from three different sight-lines, to probe how well MAGPHYS can recover key galaxy properties based on 21 bands of UVfar-infrared model photometry. MAGPHYS yields statistically acceptable fits to > 99 per cent of the pixels within the r-band effective radius and between 59 and 77 per cent of pixels within 20 kpc of the nucleus. MAGPHYS is able to recover the distribution of stellar mass, star formation rate (SFR), specific SFR, dust luminosity, dust mass, and V -band attenuation reasonably well, especially when the pixel size is 1 kpc, whereas non-standard outputs (stellar metallicity and mass-weighted age) are recovered less well. Accurate recovery is more challenging in the smallest sub-regions of the disc (pixel scale 1 kpc), where the energy balance criterion becomes increasingly incorrect. Estimating integrated galaxy properties by summing the recovered pixel values, the true integrated values of all parameters considered except metallicity and age are well recovered at all spatial resolutions, ranging from 0.2 kpc to integrating across the disc, albeit with some evidence for resolution-dependent biases. These results must be considered when attempting to analyse the structure of real galaxies with actual observational data, for which the 'ground truth' is unknown.