This research aimed to reduce the variability on the data obtained from differential scanning calorimetric (DSC) analysis of the isothermal crystallization kinetics of cocoa butter. To enable transformation of the DSC crystallization peak to a sigmoid crystallization curve, the DSC peak area has to be integrated. Usually, the start and end points of the crystallization peak are determined visually. The result of this visual determination appeared to be very much dependent on the operator, but also differed considerably when the same operator performed the integration several times. By proposing an objective calculation algorithm to determine the start and end points of integration, the variability caused by the operator during the integration procedure could be eliminated. Furthermore, sample preparation and the DSC heating protocol to melt the sample prior to crystallization were studied. Three heating protocols (65 degreesC for 15 min, 65 degreesC for 30 min and 80 degreesC for 15 min) were compared and it was shown that holding at 65 degreesC for 15 min was sufficient to eliminate any influence of sample history. Two different sample preparation procedures were compared and it appeared that a change in sample preparation procedure had a significant influence on the measured crystallization process. It is thus important to keep this method constant to eliminate the variability caused by it.