Objective: To investigate risk factors for frequent attendance in primary care over time, contrasting models based on baseline and time-varying characteristics. Methods: Analysis of data from the Personality and Total Health (PATH) Through Life Project: a representative community cohort study from the Canberra region of Australia. A balanced sample of 1734 respondents, initially aged in their early 40s, were assessed on three occasions over 8 years. The survey assessed respondents' experience of chronic physical conditions, self-reported health, depression symptoms, personality, life events, socio-demographic characteristics and self-reported medication use. Survey data were linked to respondent's own administrative health service use data, and used to generate an objective measure of general practitioner (GP) consultations over a 12-month period. For each gender, respondents in the (approximate) highest decile of GP consultations at each time point were defined as frequent attenders (FAs). Results: Analysis showed chronic health conditions, self- reported health, mental health and medication use measured at baseline was associated with FA status, with some gender differences evident. However taking into account of changing circumstances improved the model fit and the prediction over FA status over time. Conclusions: The study showed that there is considerable variability in frequent attender status over the study period. While baseline characteristics can predict current and future frequent attender status, it is clear that frequent attender in primary care does reflect changing circumstances over time.