In different fields of science and engineering (medicine, economy, oceanographic, biologic systems, etc) the False Nearest Neighbors (FNN) method has a special relevance. In some of these applications, it is important to provide the results in a reasonable time, thus the execution time of the FNN method has to be reduced. This paper describes a parallel implementation of the FNN method for distributed memory architectures based on kd-tree. A "Single-Program, Multiple Data" (SPMD) paradigm is employed using a tree decomposition approach where each processor runs the same program but computes a different sub-tree called local tree. As far as the authors know, there is not any parallel implementation of the FNN method based on kd-tree, consisting this implementation the main contribution of the paper. The accuracy and performance of the parallel approach are then assessed and compared to the best sequential kd-tree based implementation of the FNN method, executing from 2 up to 64 processors and running a Lorenz time series and an electrocardiogram signal as case studies. Results are discussed in terms on execution time, speed-up, and efficiency. In terms of speed, our approach was 3 similar to 20 times faster than sequential algorithm. (C) 2010 Published by Elsevier Ltd.