<p>Evolutionary biologists commonly use landmark-based geometric morphometrics to characterize and compare biological shapes. Advances in three-dimensional imaging hold potential to catapult morphometric studies, but quantifying shape from 3D surfaces efficiently remains challenging. Several surface-only morphometric procedures have been proposed, which align surfaces to one another and use their differences as the basis of shape quantification and analysis. Comparisons of landmark-based and surface-only methods for specific datasets have revealed mixed results, but these studies utilized empirical datasets where the true patterns of shape differences are unknown. In this study we performed a simulation experiment where the true shape differences between objects were known, and evaluated the performance of two landmark-based and four surface-only methodologies under these conditions. Several procedures, including landmarks plus semilandmarks, as well as the surface-only method DAA (and to a lesser extent auto3d), most consistently recovered the known patterns of shape variation. Results from other surface-only procedures (GPSA and pseudoLM) were more variable. Notably, patterns in more complex shapes were better characterized than patterns in simpler shapes, implying that automated surface-only methods were challenged when presented with shapes whose principal axes were similar, or where structures contained few localized shape differences for these methods to properly orient the surfaces. Overall, our findings highlight that while biologically informed landmarking remains imperative as a foundation for morphometric analyses, some surface-only procedures do offer a powerful and efficient means of characterizing shape, and hold promise for morphometric studies.</p>