By using the data of the visibility observations and NCEP reanalysis data during 2002 － 2011，the processes of low visibility weather in North China has been studied．The results show that the circulation situation on 500 hPa that caused low visibility weather in studied area could be divided into three types including two troughs and a ridge type，low trough type and zonal flow type，and the automatic identification system could be established to recognize the three weather types and eliminate other situations from these three types． On this basis，the use of K － index，the dew point temperature difference and pseudo－equivalent potential temperature difference between 500 hPa and 850 hPa gave a further physical diagnosis，and at last the forecast of low visibility weather could be obtained． The low visibility weathers occurring in North China in 2012 were forecasted by using the method mentioned above，the results show that the prediction effect of this forecast method was fine．