In　this　paper,　we　introduce　the　D-p-metric　of　set-valued　random　variables.　We　prove　that　the　Dp-metric　space　of　set-valued　random　variables　is　complete,　and　represent　the　equivalent　propositions　of　convergence　for　set-valued　random　variables　with　respect　to　D-p-metric.　Finally,　we　give　several　definitions　of　the　variance,　covariance　and　correlation　coefficient　of　set-valued　random　variables　by　the　D-p-metric,　and　discuss　some　useful　results.