It is common belief that in Raman mapping, the sampling depth from which Raman signal is collected originates from a probe laser spot size of approximately 1–5 µm3. Actually, the active pharmaceutical ingredient (API) crystals detected by mapping on the sample surface distribute from surface to over a few tens to hundreds of microns into the sample, as determined by z-slices mapping in this context. It was also found that the score of API crystals detected in a chemical image decrease with their depth into the sample. Therefore, a larger threshold of the binary chemical images can be used to restrict the sampling depth and consequently eliminate the problem of ‘saturation’ by quantifying merely the amount of API crystals within a sampling volume equivalent to a smaller sampling depth rather than the overall detected crystalline API in the chemical images. Eventually, one single linear quantification calibration was established over the crystallinity from sub-percent (<1%) to over 20% w/w of a model sample by using a simple univariate analysis of binary chemical images method. Copyright © 2013 John Wiley & Sons, Ltd.