This paper proposes a weighted fuzzy C-means (W-FPCM) clustering algorithm. It is based on the fuzzy possibilistic C-means (FPCM) algorithm. The idea of W-FPCM came from the Pareto principle. W-FPCM associates different weights to variables when computing distance in the process of clustering after filtering out less important variables. The algorithm performs well for data sets from UCI (University of California, Irvine) in terms of three different evaluation methods. The first is based on accuracy, the second is a refinement of the FPCM's objective function; the third is Kosko's fuzzy entropy formula. The main difference between the conventional feature selection fuzzy clustering algorithms and ours is that our weighting scheme runs through out the clustering process while the others just for selection of variables.