In view of the low positioning accuracy brought about by RSS similarity differences and spatial ambiguity in WiFi indoor positioning, a constrained adaptive K-nearest neighbor indoor positioning method has thus been proposed, which integrates weighted RSS similarity and Spearman correlation coefficients. First, different weight values are assigned based on varying RSS values, with the Spearman correlation coefficient combined to obtain an adjusted weighted RSS Euclidean distance for calculating the similarity between online location fingerprints and those in the database. Then, the number of candidate reference points can be obtained based on the adaptive target population screening. Finally, the final reference point is selected by setting the rectangular window and downward parameter based on the prior position, with the final position estimated as their weighted center. The experimental results show that the proposed algorithm can effectively improve the positioning accuracy compared to other algorithms, with a 52.42% improvement compared to the WKNN algorithm.