The regional convergence hypothesis asserts that regional differences in the United States have declined over the past century. Convergence has been documented for many demographic and socioeconomic variables; but it is less clear that regional cultural differences have declined. This paper examines the convergence hypothesis through the variable of religious affiliation. Three hypotheses are tested: that regional differences have declined in white Protestant proportions, in denominational proportions within white Protestantism, and in Catholic proportions. Results for all three tests reveal that, contrary to the predictions of the convergence hypothesis, regional differences in religious affiliation in the United States have grown considerably stronger during the twentieth century. The observed patterns of divergence were brought about by spatiaI variations in denominational growth, which tended to reinforce earIier regional differences. These results suggest that regional prominence has been an important factor in denominational growth this centuu.It has often been asserted that regional cultural differences within American society have been weakened over the past century, by a variety of homogenizing forces: interregional migration, urbanization, the mass media, a nationally integrated economy, and advances in communication and transportation technology. Many studies have demonstrated that the various regions of the United States have converged since the mid-nineteenth century with respect to socioeconomic and demographic variables such as fertility, education, income, voting behavior, sex ratio and racial composition (Easterlin, 1960;Labovitz, 1962Labovitz, , 1965McKinney and Bourque, 1971;Sharpless, 1980). It is less clear, however, that cultural differences have declined; the hypothesis that cultural convergence would accompany economic and demographic convergence (Labovitz and Purdy, 1970) has not been convincingly confirmed or refuted by empirical studies. Glenn and Simmons (1967), for example, did not find a general decrease in regional differences in their analysis of opinion survey data pertaining to religion, race relations, work, morals and politics. On the contrary, regional differences were increasing for many of the attitudes examined. Reed (1972) also found no indication of convergence between the South and the rest of the country using similar data on three themes associated with Southern distinctiveness: religion, violence and localism. While convincing for the period of time they cover, the 1950s and 1960s, the results of these studies are limited by their narrow temporal scope. Analyses of the convergence ofeconomic and demographic variables indicate that the process has involved long-term changes extending over at least a century, and is characterized by considerable short-term variation (Labovitz and Purdy, 1970;Sharpless, 1980). Analysis of trends over one or two decades could easily reveal stability in regional differences, of even an increase, while a more extended analysis would not. ...