Long-term intensive use of mineral fertilizers in double rice-cropping systems has led to soil acidification and soil degradation. Manure fertilization was suggested as an alternative strategy to mitigate soil degradation. However, the effects of long-term mineral and manure fertilization on rice grain yield, yield stability, soil organic carbon (SOC) content, soil total nitrogen (TN) content, and the underlying mechanisms are unclear. Based on a long-term experiment established in 1981 in southern China, we compared four treatments: no fertilizer application (Control); application of nitrogen–phosphorus–potassium (NPK); NPK plus green manure in early rice (M1); and M1 plus farmyard manure in late rice and rice straw return in winter (M2). Our results showed that 37 years of NPK, M1, and M2 significantly increased rice grain yield by 54%, 46%, and 72%, and yield stability by 22%, 17%, and 9%, respectively. M1 and M2 significantly increased SOC content by 39% and 23% compared to Control, respectively, whereas there was no difference between Control and NPK. Regarding soil TN content, it was significantly increased by 8%, 46%, and 20% by NPK, M1, and M2, respectively. In addition, M2 significantly increased bacterial OTU richness by 68%, Chao1 index by 79%, and altered the bacterial community composition. Changes in soil nutrient availability and bacterial Simpson index were positively correlated with the changes in grain yield, while shifts in bacterial community were closely related to yield stability. This study provides pioneer comprehensive assessments of the simultaneous responses of grain yield, yield stability, SOC and TN content, nutrient availability, and bacterial community composition to long-term mineral and manure fertilization in a double rice-cropping system. Altogether, this study spanning nearly four decades provides new perspectives for developing sustainable yet intensive rice cultivation to meet growing global demands.