<p>Many acquisition functions are available to improve active learning-based kriging models while conducting reliability-based design optimization (RBDO). A considerable challenge for computationally expensive models is deciding which acquisition function provides the greatest chance to complete the optimization with a minimum number of function evaluations. This paper presents a comprehensive comparative study of nine different acquisition functions in terms of the number of completed optimizations, total function evaluations, and repeatability. The comparative study was conducted. The comparative study was conducted on problems with varying levels of input uncertainty and surrogate uncertainty thresholds to evaluate the performance across a range of problem settings. Two well-known mathematical examples and one engineering example are employed to compare the performance of different acquisition functions. A unified metric is proposed to evaluate the overall performance of different acquisition functions for RBDO. The results of the comparative study show that: (1) The performance of the acquisition functions can be categorized into two distinct groups based on whether they include a term for the joint probability density function; the acquisition functions within a group have similar performance; and the acquisition functions that included a term for the joint probability density function had the best performance. (2) Local approximations have a higher success rate of finding the RBDO optimum than global approximations due to higher surrogate model fidelity in the optimum region. (3) This paper also explores a common typographical error in the expected feasibility function (EFF) that limits the ability of the function to explore the design space. This error decreases the effectiveness of EFF when compared to other acquisition functions. </p>