Objectives The effects of different amounts of potassium (K) fertilizer on rice yield, K utilization efficiency and the capacity of soil K supply under straw return in rice production region of northeast China were systematically studied, in order to provide scientific basis for rational K fertilizer application under straw return of rice.

Methods A located field experiment was conducted in Qianguo County of Jilin Province during 2015–2019. Six rates of K fertilizer (K₂O) treatments were setup, including 0 kg/hm² (K0), 30 kg/hm² (K30), 60 kg/hm² (K60), 90 kg/hm² (K90), 120 kg/hm² (K120) and 150 kg/hm² (K150). After harvest each year, the grain and biomass yield, K content in plant were measured, and the contents of soil readily available, slowly available and total K in 0–20 cm and 20–40 cm depth were analyzed. The K accumulation, K utilization efficiency and apparent K balance in soil were calculated.

Results Compared with K0 treatment, the grain and biomass yields were averagely increased by 7.6%–14.5% and 6.3%–10.9%, with the highest yields in K60 and K90 treatments. The harvest indexes were not significantly different among the K treatments. K recovery efficiency, agronomic efficiency and partial productivity declined with increasing of K application rate. The treatments of K60, K90, K120 and K150 increased the contents of readily available K and slowly available K in 0–40 cm soil, but did not in the total K contents, compared with K0 and K30 treatment. There was no significant difference in readily available K and slowly available K in 0–40 cm soil among the treatments of K90, K120 and K150. The apparent K balance of soil were in deficiency under K0 and K30 treatments during the five-years period, just in balance under the K60 treatment, and surplus under the K90, K120 and K150 treatments. When surplus rate of K was 0, the simulated K application rate was 53.1 kg/hm², the achieved grain yield would be 10035 kg/hm², readily available K contents in 0–20 cm and 20–40 cm soil be 103.04 mg/kg and 91.56 mg/kg, and the K recovery efficiency, agronomic efficiency and partial productivity be 40.4%, 21.2 kg/kg and 202.2 kg/kg, by simulating between K fertilizer rate, grain yield, soil readily available K content, K utilization efficiency and surplus rate, respectively.

Conclusions Under the tested condition, the soil K apparent balance will be in deficit when annually K₂O rates is less than 30 kg/hm², just in balance when K₂O 60 kg/hm² is applied, and in surplus when K₂O rate is higher than 90 kg/hm². The high rice yield will be achieved at K₂O rate of 60–90 kg/hm², and the soil readily available and slowly available K contents will not be further increased at K₂O rate higher than 90 kg/hm². The simulated application rate with annual soil K surplus of zero is at the range of K₂O 50–56 kg/hm², in which the rice yield, the K fertilizer utilization efficiency and the capacity of soil K supply will be the highest under this experimental condition.