The basis for experiment was two-person redistribution game played by subjects who had previously earned 15, 30, or 60 PLN in proportion to the job actually done. Subjects were paired to form dyad societies in which income inequality occurred. Single stage of the game comprised two kinds of decisions. First both members of the dyad society were to democratically establish the rate of redistributive income tax which benefited the poorer at the expense of the richer. Then (after fiscal mechanism had been executed) they could pass freely any amount of money to each other. Thus the game allowed both for involuntary transfers brought about by fiscal system, and voluntary transfers left to players’ individual discretion. While voluntary transfers were costless (there’s no need to employ resources to force anyone to do what one is willing to do on one’s own), involuntary transfers entailed extra cost defined as a percentage of total tax revenues. Half games were played under 10% cost, and half under 30% cost – the latter one made a particularly sharp case of Okun’s leaky bucket. Due to the inefficiency of tax transfers, it was socially desirable to substitute voluntary transfers for costly involuntary ones thus reducing loss-generating tax-rate to zero. Though in each game there was a range of Pareto-optimal substitutions, individual rationality urged poor player to demand maximal tax, and rich player to refuse any free transfer. That feature of the redistribution game gives it a semblance of asymmetric prisoner’s dilemma with continuous strategy sets. Results of the experiment show that subjects’ behavior largely depended on income differences within the dyad while adverse impact of fiscal cost on efficiency was hardly taken into account. Certain measure of crowding-out was present. To much surprise, in later games structural aspects of the situation were almost completely overridden by historical factors.