To investigate the evolution and significance of Q waves in type 2 diabetes, we studied 36 patients of mean (+/-S.D.) age 69.9+/-7.1 years from the longitudinal observational Fremantle Diabetes Study (FDS). All had (i) neither history/symptoms of coronary heart disease (CHD) nor pathological Q waves at FDS recruitment between 1993 and 1996, (ii) five consecutive annual assessments by FDS close-out in 2001, and (iii) contrast-enhanced cardiac magnetic resonance imaging in 2005. At this latter assessment, there were (i) 9 with no history of CHD or Q waves during follow-up (Group 1), (ii) 13 with Q waves on >/=1 electrocardiogram but no CHD history/symptoms (Group 2), and (iii) 14 with CHD history/symptoms irrespective of electrocardiographic status (Group 3). Of 20 episodes of new Q waves in 17 Group 2 or Group 3 patients during FDS follow-up, 17 (85%) resolved within 2 years. A myocardial infarction (MI) was detected by CMR in three patients (8.3%; one subendocardial in Groups 1 and 3, one non-full-thickness in Group 3) but these did not correlate with electrocardiographic appearances. Q waves may have unreliable pathological significance in type 2 diabetes, including as a marker of silent MI.