The aim of this paper was to clarify the relationship between subjective preference for air-conditioner sounds and sound quality indices, such as factors extracted from the autocorrelation function (ACF) inside a car in summer and winter. We evaluated subjective preference using a paired comparison method. Specifically, we performed multiple regression analyses with scale values of preference as the outcome variable using a linear combination of LAeq, the ACF factors, and their standard deviations. The results indicated that total subjective preference for air-conditioner sound can be predicted using LAeq, the amplitude of the first maximum peak of the ACF, φ1, which corresponds to the pitch strength, and the width of the first decay of the ACF, Wφ(0), which corresponds to the spectral centroid. We found preference to increase with a decrease in LAeq and φ1, and an increase in Wφ(0). Thus, air-conditioner sounds with quieter levels, weaker pitch strength, and lower spectral centroid were associated with higher levels of preference. The temperature of the atmosphere had no effect on subjective preference for air-conditioner sounds.