OBJECTIVES: Loneliness among elderly has been associated with poor health outcomes. Studies have shown that a significant proportion of elderly are affected by loneliness, yet literature addressing loneliness as an independent predictor of healthcare utilization are few. Our study aimed to determine the association between changes in loneliness status across two years and prospective physician visits among community-dwelling elderly in Singapore. METHODS:   We obtained data from a nationally representative longitudinal survey of community dwelling Singaporeans aged 60 years and older. Out of 3103 responses, we excluded proxy interviews (n = 365; 11.8%), resulting in a final analysis cohort of 2738 respondents. Using the number of physician visits in the past 30 days as dependent variable, we performed negative binomial hurdle regression controlling for predisposing characteristics, enabling factors, needs, social network (within and without household) and change in loneliness status between two waves of the survey. We hypothesized that chronic loneliness over the two year period would be associated with any and/or fewer prospective (Wave 2) physician visits. RESULTS: Approximately 23% respondents suffered from chronic loneliness; a further 19% developed loneliness whereas 33% recovered from loneliness by Wave 2. Chronic (OR 0.75 ± 0.09) and recently-developed loneliness (OR 0.71 ± 0.08) were significantly associated with lower odds of physician visits, compared to being never lonely in both surveys. Meanwhile, those who recovered from loneliness and visited their physicians were significantly associated with lower number of visits (-0.71 ± 0.22). CONCLUSIONS: Our study found significant association between loneliness and lower odds of physician visits regardless of whether the loneliness was recently developed or chronic. In addition, respondents who had loneliness in the past also had significantly lower number of visits to physicians compared to those who were never lonely, if they did utilize physicians. Further studies on the underlying behavior and the health consequences are warranted.