In experiments, there are usually two general ways of obtaining dominants and subordinates to test for the effect of recent experience upon ulterior behavior and dominance. One is to 'impose' such an experience on the contestants by a priori deciding which individual of the pair will become the dominant and which will become the subordinate through the use of rigged contests. The second technique is to let contestants 'self-select' the winner and loser by waiting for the spontaneous outcome of dyadic encounters between two usually well matched opponents. These two techniques of obtaining dominants and subordinates probably represent extreme cases on a single continuum of investment made by animals to settle dominance. To test this, we compared dominants and subordinates obtained from these two techniques in Xiphophorus fish males. It was found that pairs obtained through rigged contest (R) were much more aggressive in subsequent encounters than pairs in which the dominant and subordinate could self-select (S). They recuperated more rapidly from handling, initiated contact earlier, took more time to assess each other, and fought for a longer period of time. Prior-winners and prior-losers of the R condition more frequently relied on aggressive behavior during contest than that of the S condition. As a consequence, prior-winners and prior-losers of the R condition won equally the subsequent contest. On the contrary, prior-winners of the S condition defeated their prior-loser opponent in a majority of cases. These results can be tentatively explained by the following principle, winning or losing against a well matched opponent would leave more 'experience' than winning over a much weaker opponent, or losing to a much stronger one. This reinforces the hypothesis that prior-experiences are not qualitative states but come in various degrees.