Strengths of this study include the large sample size and the representativeness of the sample. However, in considering the results of this study, several limitations must be kept in mind. First, all data are based on respondents' reports of whether they were advised to lose weight and thus are subject to recall bias. Advice to lose weight may be brief and therefore may not be recalled. Published data on recall of physician advice to modify behaviors is limited to smoking cessation counseling. Folsom and Grimm24 found that only 60% of patients who received smoking cessation counseling as part of a research protocol recalled such advice 3 months later. Physicians' recall, however, might not be as accurate as patient recall of advice; Pbert and colleagues25 found that physicians in a smoking cessation counseling study advised only 71% to quit (validated by tape recording), although 100% of physicians reported dispensing such advice. With longer delays, there may be greater problems with recall. Frank and colleagues26 found that 9.1% of individuals who initially recalled receiving advice to quit smoking no longer recalled receiving such advice several years later. Finally, the opposite problem, namely, overreporting of advice, might occur. Ward and Sanson-Fisher27 compared audiotapes with patient exit interview reports and found that patients overreported receiving advice to quit smoking (specificity, 82%; sensitivity, 92%). Because of social pressure, this overreporting of advice may be particularly apparent in individuals with the highest BMIs. Changes in respondents' BMI between the physician visit and the time of the BRFSS survey might also have occurred. This effect is probably small, however, because weight changes over time average 0.45 to 0.9 kg per year.28,29 Finally, the outcome question is worded, "Has a doctor, nurse, or other health professional given you advice about your weight?" It is possible that patients are recalling advice from another health professional (eg, nutritionist or dietitian), and conclusions might not apply specifically to physicians, although, as the task of providing health advice typically falls to the physician, we are confident that the results are applicable to physicians. More research is needed to understand the frequency of counseling from other health care providers because, in research settings, other health care providers can be effective at encouraging behavior change.