With the increasing use of the object-oriented database software, system performance has become an important issue in the system procurements and evaluation process. It is vital to understand and predict the functionality and performance characteristics of the systems to be procured in the real world settings so as to justify the use of the object-oriented systems over the traditional systems. Benchmarks are considered the most common approach to test and measure the quantitative performance of database systems. We have seen a set of de facto standard benchmarks such as the OO1, HyperModel, ACOB, and OO7 benchmarks receiving more and more attention and acceptance. It is hence the intent of this paper to provide an elaborate version of analysis of these benchmarks and to give a systematic comparison of these standards with a multi-set of comparison criteria. Through the comparison and contrast, we point out the essence of features of a desirable database benchmark.