In　this　article,　we　study　model　selection　and　model　averaging　in　quantile　regression.　Under　general　conditions,　we　develop　a　focused　information　criterion　and　a　frequentist　model　average　estimator　for　the　parameters　in　quantile　regression　model,　and　examine　their　theoretical　properties.　The　new　procedures　provide　a　robust　alternative　to　the　least　squares　method　or　likelihood　method,　and　a　major　advantage　of　the　proposed　procedures　is　that　when　the　variance　of　random　error　is　infinite,　the　proposed　procedure　works　beautifully　while　the　least　squares　method　breaks　down.　A　simulation　study　and　a　real　data　example　are　presented　to　show　that　the　proposed　method　performs　well　with　a　finite　sample　and　is　easy　to　use　in　practice.