The　Models-3　Community　Multiscale　Air　Quality　(CMAQ)　modeling　system　was　integrated　with　the　meteorological　simulation　tools,　including　(a)　the　Fifth-Generation　NCAR　/　Penn　State　Mesoscale　Model　(MM5),　and　(b)　the　Advanced　Regional　Prediction　System　(ARPS).　The　integrated　system　was　then　applied　to　Beijing　region　to　study　the　PM10　concentrations　in　Jan　2002.　Most　updated　and　detailed　emission　inventories　along　with　large　amount　of　original　pollutant　source　data　ensured　considerable　prediction　accuracy.　The　nesting　technique　used　in　the　coupled　modeling　system　provided　the　inner-most　simulation　domain　with　reasonable　boundary　conditions.　The　simulation　outputs　were　compared　with　ground　level　observations　from　local　monitoring　stations　and　lidar　records　reflecting　variation　of　PM10　concentration　with　height.　It　was　found　that　the　modeling　system　offered　reasonable　prediction　accuracy,　with　most　observed　key　features　being　captured.　Examination　of　the　high　episode　of　PM10　indicated　that　the　pollution　in　Beijing　was　not　only　from　local　emissions　but　also　from　regional　atmospheric　transport.　The　wind　field　also　played　a　key　role　in　the　dispersion　of　pollutants.　This　study　represents　a　first　attempt　to　combine　the　meteorological　models　of　MM5　and　ARPS　with　CMAQ　to　effectively　account　for　small-scale　meteorological　characteristics　and　complex　lower　boundary　conditions　in　a　semi-basin　region.　©　2005　ISEIS　-　International　Society　for　Environmental　Information　Sciences.