Abstract: The evaluation of source rocks from well loggings can make up for the insufficiency of source rock core and discontinuity of samples examined.Organic-rich source rocks show relative increase in response to resistivity,neutron,sonic transit time and gamma loggings and decrease in response to density logs.Based on this response relationship,well logging data not only can be used to identify qualitatively source rocks from non-source rocks,but also predict quantificationally organic carbon content in source rocks.The predicting models proposed during past fifty years can be summed up as qualitative identification model,volume model,cross plot model,regression equation model,ΔlogR technique,and fuzzy model.The TOC of Wenchang Formation source rock in Huizhou Depression and Panyu 4 Sag in the Pearl River Mouth Basin shows similar degree of correlation respectively to resistivity,neutron,sonic transit time,gamma and density.So,the five variate regression equation should be the best prediction model for this study area.The organic carbon content of whole Wenchang Formation of one well in Panyu 4 sag predicted by this numerical model,and it shows good results verified by the lithological description and the TOC test of core samples in this well.