Abstract: Pyrite is commonly found in non−ferrous metal sulfide ores, while the oxidation−dissolution reactions with dissolved oxygen in the pulp significantly affect its floatability. However, previous studies mainly focused on the influence of surface oxidation behavior on flotation behavior, while neglecting the effects of ionic dissolution. To address this gap, this study investigated the oxidation−dissolution characteristics and floatability variations of pyrite, and employed BP neural network to analyze the correlation between recovery rates and released ions. The oxidation−dissolution experiments reveal that the oxidation degree of pyrite increased progressively over time during flotation, accompanied by the increased concentrations of Fe²⁺, Fe³⁺, and SO₄²⁻, reaching up to 68.01 mg/L, 20.44 mg/L, and 107.02 mg/L, respectively. Correlation coefficient analysis demonstrates a significant negative correlation between pyrite recovery rates and Fe²⁺/Fe³⁺ concentrations in the pulp (r₁= −0.42; r₂= −0.84), with Fe³⁺ exhibiting the strongest negative correlation. This phenomenon can be attributed to the dissolution of iron ions covering surface−active adsorption sites, which reduces the adsorption capacity of butyl xanthate and ultimately leads to decreased recovery. Furthermore, the BP neural network model demonstrates robust performance in predicting pyrite recovery, achieving a coefficient of determination (R²) of 0.9988, a mean squared error (MSE) of 0.5316, a mean absolute error (MAE) of 0.0063, and a relative average error of only 2.47%. These findings improve our understanding of pyrite flotation behavior and provide valuable insights into studying oxidation−dissolution reactions in other sulfide minerals.