The fear of crime is a politically important measure. Much is made of year to year changes in levels. Data were from the first two waves of the Community Living and Integration Survey, conducted by the Psychological Research Center, Faculty of Social Sciences, University of West Indies, St. Augustine, Trinidad. The first wave was conducted in September 1999 and the second in September 2000. The sampling frame was a multistage cluster design. The results show that the effects of differing question order cannot lightly be dismissed. The safety/afraid questions were asked as questions number 17-20 and 29-42 in the first wave of the Trinidadian survey, but as questions number 2-5 and 6-19 in the second wave questionnaire. A number of respondents gave different answers to the same questions after 1 year. Many things can change in people’s lives in the space of 12 months, and this may lead some to feel more secure, and others less so. It may well be the case that respondents gave little thought to their responses. The social psychological concept of attitude ambivalence, the fluctuating assessment of people, objects, and issues, can assist in explaining the individual inconsistencies over time. Attitude inconsistency may even exist at a deep level. Inconsistency between behavior and attitudes may be the result of being unaware of particular attitudes or feelings toward an issue; situational constraints preventing one from expressing an attitude; impression management; attitude strength; accessibility of the attitude to an individual’s mind; and importance and vested interest of the attitude in question. There are three possibilities for this problem: (1) these global and specific questions are an unstable measure of a stable phenomenon (the fear of crime); (2) they are a stable measure of an unstable phenomenon; and (3) they are an unstable measure of an unstable phenomenon. These results question either or both of the stability of the fear of crime, and of measures used to assess it. 11 tables, 5 notes, 42 references