Starch conversion into sugars is one of the most important indicators for predicting the optimal harvest dates for apples. The starch conversion is usually assessed using an iodine test, in which cut fruit are dipped in iodine solution and stain patterns are rated. A methodology based on the image analysis of iodine-treated apple slices was used to quantify the uncertainty on the traditional starch index measurements. Considering the subjective interpretation of different inspectors, the errors for certain image assignments were up to 60%. The averaged values of index assignments of groups of images, however, were for most starch indices not significantly different. It was noticed that the subjective interpretation was related to the high number of indices on the chart, the non-linear scale and a sequence disorder of the reference chart. A modified logistic model was proposed to represent the sigmoidal starch index pattern during the preharvest period by two model parameters. Based on these two parameters, and the recommended starch index at harvest, the optimal harvest date was determined. With this technique, the orchard and cultivar effects were analysed.