Clustering is proposed to cluster the high dimensional data, into clusters of data that exhibit some similarities. Due to this ability, it has been chosen to solve many problems in various areas, including tropical wood species classification. It has ease the recognition process which has been done manually before. Pheromone-based Kohonen Self-Organizing Map (PKSOM) is proposed a clustering tool to cluster the wood datasets; the filtered and raw datasets. This paper discusses the performance and scalability of modified Kohonen Self-Organizing Map (KSOM), named Pheromone-based KSOM (PKSOM). The PKSOM algorithm is trained and tested using two types of wood datasets; the raw wood dataset and filtered dataset using Genetic Algorithm (GA). The results are then being compared with two conventional clustering methods; KSOM and standard K-Mean. As a conclusion, PKSOM has produced 97.5% of accuracy for the raw wood dataset and 97.55% for the filtered wood dataset, slightly higher compared to the other two methods.