Phylogenetics plays a pivotal role in evolutionary biology by elucidating the relationships among species through the analysis of DNA and protein sequences. Phylogenetic trees constructed from such analyses provide critical insights into evolutionary pathways, ancestral connections, and biological diversity. In this study, we conducted a systematic evaluation of five widely used phylogenetic inference methods: Unweighted Pair Group Method with Arithmetic Mean (UPGMA), Neighbor-Joining (NJ), Minimum Evolution (ME), Maximum Parsimony (MP), and Maximum Likelihood (ML). To ensure robust assessment, we generated simulated datasets under diverse conditions, including variations in the number of taxa, sequence lengths, and insertion-deletion rates. Our evaluation framework followed an inverse problem approach, where reference trees were first simulated, and data was generated from these trees. The generated data was subsequently used to infer new phylogenetic trees, which were then compared to their corresponding reference trees. Performance was assessed using two key metrics: the Robinson-Foulds Distance (RFD) for topological accuracy and the Mean Branch Length Distance (MBLD) for evolutionary distance accuracy. Results revealed substantial differences among methods, with UPGMA emerging as the most consistent performer in both topological structure and branch-length estimation. Notably, UPGMA achieved superior performance despite the simulated trees not adhering to a strict molecular clock, which makes this finding particularly surprising.