Abstract: 【Objectives】Shifting part of nitrogen fertilization from the basal-tillering to panicle initiation (PBSP) of rice is beneficial to improve nitrogen use efficiency and grain yield of paddy rice, however, limited information is available on the yield performance among different hybrid rice cultivars grown under water-logged paddy field. This study was conducted to determine cultivar difference in grain yield and in the response of source-sink structure to PBSP. 【Methods】 Twenty hybrid rice cultivars were grown under the same total nitrogen (N) fertilizer input with two different application strategies: part of nitrogen fertilizershifted from the basal-tillering stage to the panicle initiation (PBSP, the ratios of 60%, 20% and 20% of total N fertilizer were applied at basal, spikelet-promoting fertilizer, and spikelet-protecting fertilizer, respectively), and a heavy basal N plus top dressing at the early tillering stage (HBET, the ratios of 70% and 30% of total N fertilizer are applied at basal, and 7 day after transplanting, respectively). The treatments were arranged in a split-plot design with the application strategy as main plots and cultivars as subplots. The differences of grain yield and yield components between PBSP and HBET were analyzed by correlation, regression and path analysis based on the mean measurements of the triplicate. 【Results】There are significant differences of grain yield and yield components between PBSP and HBET with the F value ranging from 8.89**to 149.08**. Averagely, the grain filling percentages and thousand grain weights of the twenty hybrid rice cultivars treated with PBSP are significant higher than those with HBET, while the panicle numbers per m2 with PBSP are significant lower, and there are no significant differences in the maximum tiller numbers per m2, spikelet numbers per panicle and ratios of spikelets per m2 to leaf area. The differences of grain yield of hybrid rice cultivars between PBSP and HBET are fluctuated. The average grain yield of hybrid rice cultivars with PBSP is 8981.90 kg/ha, which is 1.62% higher than that with HBET. Four of the twenty hybrid rice cultivars (nei 5you306, Rong18you447, Nei5you317 and Chuanguyou7329) produce significant higher grain yields under PBSP. For a given hybrid, negative correlation is observed between grain yield differences of the PBSP vs the HBET and the spikelet numbers per panicle (r=-0.7870** for PBSP, and -0.7986** for HBET). The relatively higher grain yields with PBSP are mainly due to the higher panicle number per m2, spikelets per panicle and grain filling percentage. The hybrid rice cultivars with larger panicles produce lower grain yield under PBSP, and lower maximum tillers, panicle number per m2 and spikelets per panicle are responsible for lower grain yield of hybrid rice cultivars with larger panicles. 【Conclusions】The hybrid rice cultivars with spikelet number per panicle less than 237 can be suggested as a criterion suitable for the PBSP fertilization in all-time waterlogged paddy fields.