This paper reports about the 2004 comparative evaluation of solvers for quantified Boolean formulas (QBFs), the second in a series of non-competitive events established with the aim of assessing the advancements in the field of QBF reasoning and related research. We evaluated sixteen solvers on a test set of about one thousand benchmarks, selected from instances submitted to the evaluation and from those available at www.qbflib.org. In the paper we present the evaluation infrastructure, from the criteria used to select the benchmarks to the hardware set up, and we show different views about the results obtained, highlighting the strength of different solvers and the relative hardness of the benchmarks included in the test set.