The Community Organization Sense of Community Scale (COSOC) is a frequently used or cited measure of the construct in community psychology and other disciplines, despite a lack of confirmation of its underlying 4-factor framework. Two studies were conducted to test the hypothesized structure of the COSOC, the potential effects of method bias on psychometric properties, and whether a revised measure (COSOC-R) yielded improved model-to-data fit. Study 1 included year 2002 data from two samples: (a) randomly selected community residents (n=724) of five cities in the United States, and (b) randomly selected organizational members (n=508) of community-organizing initiatives in the same five US cities. Study 2 included year 2006 data from organizational members (n=151) of community-based prevention partnerships located in the midwestern United States. Results from both samples in Study 1 confirmed that method bias from the mixed use of positively and negatively worded items had a detrimental effect on the factor structure of the original COSOC. Results of Study 2 strongly supported the hypothesized 4-factor structure of the COSOC-R (i.e., relationship to organization, organization as mediator, influence of the organization, and bond to community). Study 2 results also showed that the overall COSOC-R and its subscales were reliable and related in expected ways with psychological empowerment, community participation, and organizational involvement. Implications of the study and strategies to further develop the COSOC are discussed. © 2008 Wiley Periodicals, Inc.