In this paper, the possibility of a bidirectional relationship between mass media use and political participation and a possible interaction of sex with other model variables are examined. A Two-Stage Least Squares analysis explores the mutual causation of mass media use and political participation, and separate analyses for male and female respondents explore the sex interactions. Data are from a 1971 survey of North Carolina. Mass media use is shown to have an effect on political participation in both the male and female subsamples. Political activity has a positive return effect on media use in the female subsample only. Results indicate that models which specify media use as a unidirectional cause of participation behavior are probably incorrect. Since empirical observations can lead to accurate inferences about the nature of a system only when the causal model is correctly specified, this analysis suggests wider use of techniques which allow the investigation of nonrecursive relationships. (Author/JM)