Description: Background: Morbidity and mortality related to Prescription Opioid Analgesics (POA) have been rising sharply in North America. Non-Medical Prescription Opioid Use (NMPOU) in the general population is a key indicator of POA-related harm, yet the role of survey question design for best NMPOU prevalence estimates is unclear, and existing NMPOU survey data for Canada are limited. Methods: We tested the impact of different NMPOU question items by comparing an item in the 2008 and 2009 (N = 2,017) samples of the CAMH Monitor surveys -- an Ontario adult general population survey -- with a newly developed item used in the 2010 (N = 2,015) sample of the CAMH Monitor survey. To control for a potential difference in the population demographics between surveys, we adjusted for gender, age, region, income, prescription opioid use, cigarette smoking, weekly binge drinking, cannabis use in the past three months, and psychological distress in our analyses. Results: The prevalence of NMPOU as measured by the 2008 and 2009 CAMH monitor (2.0% [95% CI: 1.2% to 2.8%]) was significantly different when compared to the prevalence of NMPOU as measured by the 2010 CAMH monitor (7.7% [95% CI: 6.3% to 9.2%]) (p 〈 0.001). This difference was also found when stratifying our analysis by sex (p 〈 0.001) and when adjusting for all potential confounding covariates. Conclusion: It is highly unlikely that the multifold NMPOU prevalence differences observed from the different survey items reflect an actual increase of NMPOU or changes in NMPOU determinants, but rather point to measurement effects. It appears that we currently do not have accurate estimates of NMPOU in the Canadian general population, even though these are needed to guide and implement targeted interventions. Given the current substantial morbidity and mortality impact of NMPOU, there is an urgent need to systematically develop, validate and standardize NMPOU items for future general population surveys in Canada.