An embodiment of the invention provides a fingerprint positioning method and device based on clustering. The method comprises the following steps: obtaining received signal strength (RSS) of a point to be positioned; obtaining initial clustering centers of a plurality of clusters from a fingerprint database, wherein the plurality of clusters are obtained by carrying out clustering on sample pointsof a sample set in the fingerprint database, each cluster comprises an initial clustering center, and the initial clustering center of each cluster is obtained by carrying out clustering based on theinformation that the mean square error of Euclidean distance between an undetermined initial clustering center and each sample point of the sample set in the fingerprint database is not greater thana first value, and the sample density of sample points in a cluster with the undetermined initial clustering center being the clustering center is larger than a second value; searching the initial clustering center, the Euclidean distance between which and the RSS of the point to be positioned is the closest, from the initial clustering centers of the plurality of clusters by utilizing the RSS ofthe point to be positioned; and based on the initial clustering center, the Euclidean distance between which and the RSS of the point to be positioned is the closest, carrying out fingerprint positioning on the RSS of the point to be positioned.