This paper offers an up-to-date review of problems in determining causal relationships in cultivation research, and considers the research rationales of various approaches with special reference to causal interpretation. It describes in turn a number of methodologies for addressing the problem and resolving it as far as this is possible. The issue of causal inference arises not only in cultivation research, however, but is basic to all media effects theories and approaches primarily at the macro-level whose main methodology rests on correlational studies (agenda-setting, spiral of silence, knowledge gap hypothesis, etc.). We therefore first discuss problems of causal interpretation in connection with the cultivation hypothesis, and then sketch in summary how these problems arise with other media effects theories. We first set out the basic features of the cultivation approach, then consider the difficulties with correlational studies and discuss alternative research designs - designs which are not original to us, but have been adapted for cultivation research. These comprise laboratory experiments, sequential studies, social studies and time-series procedures. Finally, we argue for multiple approaches that complement one another's advantages and balance out their disadvantages.