Sand production along with the production of hydrocarbons is a common problem in the petroleum industry. With a wide range of difficulties posed by the sanding, sanding analysis is needed to be done, therefore sanding can be delayed or even prevented. Sanding caused stresses around the wellbore exceeds the effective strength of rocks. In this thesis, onset sanding prediction is done on the perforation tunnel based on prediction methods that have been developed by Wilson (2002). The prediction was made on 3 pieces of wells that GH-2, GD-5 and SD-5. Horizontal stress changes due to reservoir depletion is a factor to consider in the prediction. Stress tensor transformation is done in order to align in-situ stress of the position of the hole perforations. Prediction results will be compared to the prediction of the beginning of sanding made by Palmer (2005) and Efrial (2006). The prediction was done and showed that there is no potential sanding in GH-2 but sanding may occur in GD-5 and SD-5 wells. The conducted analysis showed a significant effect of perforation tunnel orientation towards sanding.