Localization is a vital foundation work in Wireless Sensor Network (WSN). Almost all of location algorithms at present need the position information of reference nodes to locate the unknown nodes. But most of algorithms assume an idealistic radio propagation model that is far from the reality. This will lead to obvious difference compared with real localization of WSN. In this paper we investigate the impact of radio irregularity on the localization algorithms performance in WSN. We introduce the Radio Irregularity Model (RIM) which is established upon empirical data. With this model, this paper analyzes the impact of radio irregularity on localization algorithms. We compare three typical coarse-grained localization algorithms: APIT, Centroid and DV-HOP in simulated realistic settings. Our experimental results show that radio irregularity has a significant impact on some main evaluation aspects of localization algorithms. Some interesting phenomena is worthy of further study.