The　prediction　of　construction　disaster　is　the　basis　of　construction　safety　assessment　and　decision-making.　Gray　prediction　is　suitable　for　such　kinds　of　system　objects　with　few　data,　short　time　and　little　fluctuation,　and　Markov　chain　theory　is　just　suitable　for　forecasting　stochastic　fluctuating　dynamic　process.　By　combining　the　advantages　of　both　gray　prediction　and　Markov　chain　theory,　a　new　gray　Markov　model　is　proposed.　Markov　prediction　is　used　to　forecast　the　fluctuation　along　the　tendency.　Finally,　the　new　model　is　applied　to　predict　the　construction　disaster　death　toll　of　the　china　from　1999　to　2009.　The　results　show　that　the　new　model　not　only　discovers　the　trend　of　the　construction　disaster　death　toll　but　also　overcomes　the　random　fluctuation　of　data　affecting　precision.　©　2011　IEEE.