The study aims to explore the spatio-temporal variation of available zinc content in tobacco-growing soil in Kunming, so as to provide scientific basis and guidance for preventing and controlling soil zinc pollution in high quality tobacco-growing areas and improving plough land quality. By GPS positioning, a total of 3347 soil samples were collected from the plough layer (0-20 cm) of core area in eight tobacco-growing counties of Kunming in 2018-2020, respectively. Based on them, we analyzed the temporal and spatial variation characteristic of soil zinc. The main driving factors of soil Zn were analyzed by using regression and boosted regression tree model. The results showed that the soil available zinc was 2.53 mg/kg, and the coefficient of variation of Zn was 89.55%, and Zn variation was high in 2020. In addition, the proportion of abundant and high grade was larger (75.83%), which was at the high and enrich level. Compared with 2019, soil Zn was relatively stable, but its mean value decreased by 0.85 mg/kg than that of 2018, and its variation increased. The distribution frequency of soil Zn in rich and extremely grades decreased by 15.01%, while the proportion of suitable level sites increased by 8.34%. In 2020, the mean Zn range of the eight counties was 1.84-4.19 mg/kg, and the coefficient of variation was 55.22%-93.37%. Meantime, the variation of Zn mean value was as followed: Xundian>Anning>Songming>Fumin>Luquan, Jinning>Yiliang>Shilin. Compared with 2019, the soil Zn content in Fumin, Lujian, Songming and Xundian decreased by 0.11-0.69 mg/kg in 2020, while them in Anning, Jining, Yiliang and Shilin were relatively stable. The soil Zn content in 2020 decreased by 0.23-1.65 mg/kg than that of 2018, and among which Fumin, Songming and Xundian decreased the most. In 2020, the amount of Zn in the rich and extremely levels decreased by 0.16%-15.03% and 5.88%-23.20% compared with that in 2019 and 2018, respectively. The decrease rate was highest in Yiliang and lowest in Songming. The Pearson’s correlation, regression fitting and boosted regression tree model revealed that the most main factor causing spatial scale difference of Zn was average annual precipitation, followed by average annual temperature, evaporation, soil type, soil texture. Meantime, the soil pH value and hydrolyze nitrogen, organic matter, available potassium and phosphorus contents were selected as driving factors. Moreover, the long-term inputs of nitrogen, potassium, phosphate fertilizer also played key roles in soil Zn change. Overall, the Zn content of tobacco-growing soil was generally high, but showed a decreasing trend in recent 2 years. In the future, we should continue to reduce nitrogen, and control the application of phosphorus, potassium, organic fertilizer containing Zn and chemical fertilizer to reduce Zn input. Furthermore, the soil pH value and acidification should be considered to ameliorate zinc form and reduce the risk of high Zn concentration in tobacco and soil pollution.