Test case production is a crucial phase in the software testing lifecycle that consumes significant time, effort, and cost. As such, it is considered an optimization problem that can be addressed using metaheuristic techniques. This study aims to identify the metaheuristic techniques and their parameters used to generate optimal test data, the Unified Modeling Language (UML) diagrams and intermediate formats employed to create test cases, as well as the databases and metrics used to evaluate the performance of these techniques. A total of 46 primary studies published between 2010 and 2023 were reviewed, selected from an initial pool of 424 articles sourced from IEEE, Springer, Elsevier, and Google Scholar. The findings indicate that both single and hybrid metaheuristic techniques have been applied for test case optimization; however, the majority of studies employed single techniques, with Genetic Algorithms being the most frequently used. Furthermore, 50% of the studies did not specify the parameters used, while those that did often lacked proper documentation and failed to address the crucial balance between exploration and exploitation factors. Moreover, most studies (35) applied individual UML diagrams, mainly activity diagrams, while only 11 studies utilized multiple UML diagrams. Additionally, Graphs were the predominant intermediate format, used in 83% of the studies, whereas formats like XML, adjacency matrices, and tree structures were rarely considered. In terms of performance evaluation, most studies (21) utilized the ATM database, while 18 studies employed simple programs. Finally, while the majority of studies focused on metrics for evaluating the effectiveness of the techniques, only a few considered metrics related to efficiency (RQ6). To address these gaps, future research should consider expert opinion surveys to identify key parameters that ensure an optimal balance between exploration and exploitation. Also, future techniques should support the generation of test cases from multiple UML diagrams. The performance of these techniques should be evaluated through comparative studies using large databases, with equal emphasis on both effectiveness and efficiency metrics.