In this work, we report on the solubility of six disubstituted anthraquinone-type disperse dyes in compressed CO(2) between (299 and 346) K and up to a pressure of 20 MPa determined by a flow method. The disperse dyes belong to a series of 1,4-bis(alkylamino)-9,10-anthraquinones, where alkyl = methyl, ethyl, propyl, 1-methylethyl, butyl, pentyl, and octyl. The results of the previously investigated propyl derivative were included for comparative discussion. Significant differences regarding the solubility behavior of the individual dyestuffs were observed under the applied experimental conditions. VIS spectroscopy was applied to determine solubility, and the accessible measuring range was limited by a maximum analyzable solubility of about 450 center dot 10(-6) mol center dot dm(-3). At any experimental conditions, 1,4-bis(1-methylethylamino)-9,10-anthraquinone with a branched alkyl group shows the highest solubility for that homologous series. For example, at p = 16.83 MPa and T= 310.0 K, the solubility of 1,4-bis(1-methylethylamino)-9,10-anthraquinone amounted to s = 418 center dot 10(-6) mol center dot dm(-3)(0.135 g center dot dm(-3)). For the dyes substituted with linear alkylamino groups, the solubility passes through a maximum that is observed with 1,4-bis(pentylamino)-9,10-anthraquinone. The solubilities of the methyl and ethyl compound are comparatively low and similar. Equilibrium concentrations of the less-polar propyl- and butyl-substituted derivative are slightly lower than for the pentyl compound, and the sequence of solubility also depends on the experimental conditions. At higher temperatures and pressures, the solubility of the butyl derivative exceeds that of the propyl derivative. Ultimately, the overall solubility again goes down for the octyl compound to values in between those for the methyl- and ethyl-substituted and the propyl- or butyl-substituted compound. Inspired by their prevalent use in the literature, we applied five randomly selected empirical density-based correlations with different numbers of adjustable parameters to our data. Surprisingly, the best performance, i.e., the best agreement between experiment and correlation, for all data sets was observed for the simple concepts, and the introduction of further parameters did not automatically imply a better correlation.