Field development studies require the knowledge of well clustering. In this paper, a novel automatic well clustering approach is proposed. This approach can be applied to recognize effective well clusters and extract strong rules among reservoir data. The proposed procedure combines a classical clustering algorithm with the genetic algorithm (GA) or the particle swarm optimization (PSO) algorithm. This modified algorithm is used to improve the stability criteria and also to reduce the dependency of final results on the initialization criteria. The results indicate that PSO is much faster than GA in both the initialization and convergence steps; therefore, the PSO algorithm is implemented for the clustering study. The production history of 24 wells of an oil reservoir is investigated, and the type of decline curve is determined. Subsequently, the modified algorithm is implemented to classify production wells. In the end, the Apriori algorithm, as an associate rule mining procedure, is applied to discover and present frequent patterns based on support, confidence, and lift factors and extract important rules among wells.Overall, we proposed a novel approach for well clustering. The proposed procedure, which leads to high stability in results, is a proper alternative for conventional supervised or unsupervised well clustering methods. The proposed procedure is beneficial to obtain reliable results for reservoir management studies.