Clustering is deemed one of the most difficult andchallenging problems in machine learning. In this paper, wepropose a multilocal search and adaptive niching-based geneticalgorithm with a consensus criterion for automatic data clustering. The proposed algorithm employs three local searchesof different features in a sophisticated manner to efficientlyexploit the decision space. Furthermore, we develop an adaptiveniching method, which can dynamically adjust its parametervalue depending on the problem instance as well as the searchprogress, and incorporate it into the proposed algorithm. Theadaptation strategy is based on a newly devised populationdiversity index, which can be used to promote both geneticdiversity and fitness. Consequently, diverged niches of high fitnesscan be formed and maintained in the population, making theapproach well-suited to effective exploration of the complexdecision space of clustering problems. The resulting algorithmhas been used to optimize a consensus clustering criterion, whichis suggested with the purpose of achieving reliable solutions. Toevaluate the proposed algorithm, we have conducted a series ofexperiments on both synthetic and real data and compared it withother reported methods. The results show that our proposed algorithm can achieve superior performance, outperforming relatedmethods.