Clay mineral quantification is an under explored area in geotechnical engineering. This is primarily due to the complexity and expensive nature of the current quantification methods. Simpler prediction models using chemical properties of the soil such as Cation Exchange Capacity (CEC), Specific Surface Area (SSA) and Total Potassium (TP) have been developed to successfully predict the percentage and dominant clay minerals in a soil. The reliability of predictions of these models still needs to be addressed. Hence, in this paper, the reliability of one such method in predicting the percentage of clay mineral Montmorillonite is evaluated. There are two objectives for this reliability analysis: one is to calculate the percentage contribution of the chemical parameters (CEC, SSA and TP) towards the prediction of Montmorillonite in the soil, and the second is to determine the probability of prediction failure which is defined as the prediction being +/- 10% of the original mineral amount. First Order Second Moment (FOSM) method is used in this analysis with a slight modification in the parameter variation. For this purpose, a total of twenty natural soils and their chemical properties are used to obtain the mean and standard deviation of CEC, SSA and TP values. Two artificially mixed soil specimens with known percentages of mineral Montmorillonite are later used for validating and assessing the probability of mineralogy prediction failures.