Evaluation of secondary pore developing zone is difficult because of its strong heterogeneity and anisotropy in carbonate karst reservoirs. So a single data cannot make an accurate and complete evaluation. Aiming at the problem, firstly, the imaging logging data was used to evaluate the secondary pore parameters of carbonate reservoirs. Then, the obscure neural network technology was utilized to establish the relationship model between the secondary pore parameters and the seismic attributes of well-side traces. Finally, the porosity distribution of secondary pores all over the working area was predicted. The prediction shows that the secondary pore developing zone by seismic-logging method is in good accordance with the high quality reservoirs drilled. Therefore, the predicted results are feasible in the area.