Abstract: A 27 years of long-term experiment was conducted to evaluate the effects of K fertilizer application on rice yield and soil K status in reddish paddy soil. There were 5 treatments, namely, CK (without fertilization), NP (N, P fertilizers), NPK (N, P and K fertilizers), NP+RS (NP plus rice straw) and NPK+RS (NPK plus rice straw) in this study. The results show that the K fertilizer applications could obviously increase grain yields of early rice and late rice. The average yields of the early rice under the NPK and NPK+RS treatments are increased by 15.2% and 10.9%, respectively, and those of the late rice are increased by 17.2% and 9.1% compared with the NP and NP+RS treatments over 27 years. The 27-year yield change trends of both the early and late rice are different in various treatments. There are negative yield change trends with time in the CK and NP treatments, while there are the positive trends in the NPK, NP+RS and NPK+RS treatments. Contents of total K, slowly available K and available K in the different soil layers under the NPK and NPK+RS treatments are higher than those of corresponding soil layers under the NP and NP+RS treatments. Values of the apparent K balance in the soil-crop system are negative in all treatments except for the NPK+RS treatment, among which the values are highest in the CK and NP treatments. Long-term rational K fertilization could increase rice yield and sustain soil K fertility.