Amorphous lactose will crystallise rapidly if its glass transition temperature is reduced below its storage temperature. This is readily achieved by storing samples at ambient temperature and a relative humidity (RH) of greater than 50%. If the sample is monitored in an isothermal microcalorimeter as it crystallises, the heat changes associated with the event can be measured; indeed this is one of the methods used to quantify the amorphous content of powders and formulations. However, variations in the calculation methods used to determine these heat changes have led to discrepancies in the values reported in the literature and frequently make comparison of data from different sources difficult. Data analysis and peak integration software allow the selection and integration of specific areas of complex traces with great reproducibility; this has led to the observation that previously ignored artefacts are in fact of sufficient magnitude to affect calculated enthalpies. In this work a number of integration methodologies have been applied to the analysis of amorphous spray-dried lactose, crystallised under 53 or 75% RH at 25 degrees C. The data allowed the selection of a standard methodology from which reproducible heat changes could be determined. The method was subsequently applied to the analysis of partially amorphous lactose samples (containing 1-100% (w/w) amorphous content) allowing the quantification limit of the technique to be established. It was found that the best approach for obtaining reproducible results was (i) to crystallise under an RH of 53%, because this slowed the crystallisation response allowing better experimental measurement and (ii) to integrate all the events occurring in the ampoule, rather than trying to select only that region corresponding to crystallisation, since it became clear that the processes occurring in the cell overlapped and could not be deconvoluted. The technique was able to detect amorphous contents as low as 1% (w/w), using this integration strategy, although it was observed that the calibration plot constructed showed a negative deviation from linearity. It is suggested that such non-ideal behaviour results from the formation of varying ratios of alpha-lactose monohydrate, anhydrous alpha-lactose and anhydrous beta-lactose.