To reasonably predict the initial corrosion time of existing reinforced concrete structures, the random field of corrosion parameters is established using Karhunen-Loève(K-L) series expansion method. Then, combined with the existing initial corrosion time prediction model, a method for predicting the initial corrosion time of existing reinforced concrete structures considering the spatial correlation of corrosion parameters is proposed. The results show that the increase of correlation length can lead to the smoother surface of the random field distribution of corrosion parameters and the smaller the overall fluctuation intensity. Moreover, increasing the correlation length can result in the increase of the range of the initial corrosion time, but the mean value of the initial corrosion time is not sensitive to the change of the correlation length. In addition, the existing methods for predicting the initial corrosion time underestimates the corrosion risk of structures. In this case, the initial corrosion time of structures considering the spatial correlation of corrosion parameters can be advanced by 41.9%, 53.2% and 63.4% respectively when 0.5, 1 and 3 times of the standard correlation length are used. The research results can provide a research basis for the durability and safety evaluation of existing reinforced concrete structures.