In this paper we examine the market reaction to goodwill impairment loss and the determinants of recognition of goodwill impairments. Positive pre-event abnormal returns and negative post-event abnormal returns are observed around the write-offs announcement. After controlling for the potential earnings announcement effects, we find that the pre-event abnormal returns are not significantly related to impairment losses. This implies that observed positive stock reaction before the announcement date is attributable to the good earnings news and managers tent to announce goodwill write-offs when earnings news is good. The post-event abnormal returns are significantly negatively related to impairment losses after controlling for earnings announcement effects. The regression results support the arguments that goodwill impairment announcements have negative impacts on stock returns and the information of goodwill write-offs does not leak before the announcement date. As to the determinants of recognition of goodwill impairments, the empirical evidence shows that goodwill impairment losses are associated with the loss of economic values of goodwill and the managers' discretionary behaviors.