Description: Non-heating palaeointensity methods are a vital tool to explore magnetic field strength variations recorded by thermally sensitive materials of both terrestrial and extraterrestrial origin. One such method is the calibrated pseudo-Thellier method in which a specimen's natural remanent magnetization is alternating field demagnetized and replaced with a laboratory induced anhysteretic remanent magnetization (as an analogue of a thermoremanent magnetization, TRM). Using a set of 56 volcanic specimens given laboratory TRMs in fields of 10–130 μT, we refine the calibration of the pseudo-Thellier method and better define the uncertainty associated with its palaeointensity estimates. Our new calibration, obtained from 32 selected specimens, resolves the issue of non-zero intercept, which is theoretically predicted, but not satisfied by any previous calibration. The range of individual specimen calibration factors, however, is relatively large, but consistent with the variability expected for SD magnetite. We explore a number of rock magnetic parameters in an attempt to identify selection thresholds for reducing the calibration scatter, but fail to find a suitable choice. We infer that our careful selection process, which incorporates more statistics then previous studies, may be largely screening out any strong rock magnetic dependence. Some subtle grain size or mineralogical dependencies, however, remain after selection, but cannot be discerned from the scatter expected for grain size variability of SD magnetite. As a consequence of the variability in the calibration factor, the uncertainty associated with pseudo-Thellier results is much larger than previously indicated. The scatter of the calibration is ~25 per cent of the mean value, which implies that, when combined with the scatter of results typically obtained from a single site, the uncertainty of averaged pseudo-Thellier results will always be 〉25 per cent. As such, pseudo-Thellier results should be complementary to, and cross-validated with results from other methods. Nevertheless, the pseudo-Thellier method remains a valuable tool for obtaining palaeointensity estimates from thermally sensitive terrestrial and extraterrestrial materials and with careful data selection and analysis can yield results that are accurate to within a factor of 4 or better.