A　geology　prediction　approach　based　on　Markov　random　process　and　Bayesian　updating　procedure　is　presented　in　this　article.　It　can　dynamically　predict　the　ground　conditions　ahead　of　tunnel　face　in　a　probabilistic　way　with　very　low　cost,　when　the　tunnel　face　advances　and　new　geological　information　is　revealed.　The　application　case　study　on　Chuangshi　Tunnel　project　shows　that　the　prediction　results　fit　the　real　grounds　condition　well,　especially　for　the　short　range　ahead　of　face.　This　approach　can　be　regarded　as　a　good　complement　to　the　geophysical　prospecting　approach　during　the　construction　of　mountain　tunnels.　©　2012　Elsevier　Ltd.