The role that time frame plays in the media agenda-setting process was studied by examining the single issue of civil rights over an extended period between 1954 and 1976. The public agenda was determined from 27 Gallup polls, conducted between 1954 and 1976, which asked respondents what they considered the most important issue facing the American people. The media agenda consisted of the number of front page stories on civil rights that appeared in the New York Times during the six months prior to each Gallup poll. Zero-order correlations between the public and media agendas indicated a slight monotonic descent moving back in time to four months prior to the interviews, followed by a dramatic drop in the fifth and sixth months prior to the polls. The cumulative effect of the first two months prior to the polls was about as high or higher than all other combinations of cumulative correlations. Partialling out prior months indicated a "recency" effect during the first month. It was found that going back beyond two months prior to the public polls did not significantly improve the media agenda's prediction of the public agenda. Thus, the study provided further evidence of a strong agenda-setting effect; and it also indicated, at least for the civil rights issue, that the optimal effect span is the four to six weeks immediately prior to fieldwork. (RL)