Abstract: The ice load is an extreme environmental load received by the polar ship during its ice navigation. The field monitoring of hull structures is an important approach to study ice loads. Generally, the strain responses caused by ice loads are directly measured by installing strain sensors on hull structures, then ice loads are reversely identified according to those responses by Influence Coefficient Matrix Method. However, when strain sensors cannot work normally, the accuracy of identification results will be difficult to be guaranteed. Firstly, the influence of measuring point failure on identification results is studied by the finite element analysis of a typical polar ship's side grillage structure. Then, according to the in-depth analysis on measured strain data of multi-purpose ice-class ship Tian'en and the finite element strain data of a typical polar ship's side grillage structure, the spatial distribution of measuring point strains is determined. Furthermore, a method for identifying ice loads based on least square fitting under the influence of failure points is proposed. Finally, the ice loads under seven typical cases are identified accurately, which greatly reduces the identification error. As a result, the effectiveness of this method is verified.