The 'good genes' hypotheses of sexual selection predict that females prefer males with strong ornaments because they are in good health and vigor and can afford the costs of the ornaments. A key assumption of this concept is that male health and vigor are useful predictors of genetic quality and hence offspring performance. We tested this prediction in wild-caught lake char (Salvelinus umbla) whose breeding coloration is known to reveal aspects of male health. We first reanalyzed results from sperm competition trials in which embryos of known parenthood had been raised singly in either a stress- or non-stress environment. Paternal coloration did not correlate with any measures of offspring performance. However, offspring growth was reduced with higher kinship coefficients between the parents. To test the robustness of these first observations, we collected a new sample of wild males and females, used their gametes in a full-factorial in vitro breeding experiment, and singly raised about 3000 embryos in either a stress- or non-stress environment (stress induced by microbes). Again, paternal coloration did not predict offspring performance, while offspring growth was reduced with higher kinship between the parents. We conclude that, in lake char, the genetic benefits of mate choice would be strongest if females could recognize and avoid genetically related males, while male breeding colors may be more relevant in intra-sexual selection.