STUDY DESIGN.: A retrospective cohort study in the general population of Switzerland. OBJECTIVE.: To investigate the course of back pain (BP) across 5 years and the impact of BP history on its incidence and recurrence. SUMMARY OF BACKGROUND DATA.: Longitudinal studies on BP performed in the general population have reported varying prevalence and incidence rates. Most studies compared two points in time with varying time periods. This study adds information about the course of BP exploring five points in time with annual intervals. METHODS.: The Swiss Household Panel is a representative population-based cohort study (N = 7799). The question analyzed in the present study asked about "bad back or lower back problems at least once a month in the last 12 months (BP)." Among 7791 persons who answered this question during the baseline survey in 1999, 3881 persons (49.8%) completed all annual follow-up surveys through 2003 and represent the study sample. In each year, the 1-year prevalence, incidence, and recurrence of BP were calculated. The course of BP was analyzed according to the number of years with BP, the constancy of BP status, and the trend of BP. For each analysis, the observed frequency was compared with expected frequencies on the basis of two theoretical models. RESULTS.: In the study sample (age 44.0 ± 15.6 years, 57.7% women), BP prevalence was 33.2% at baseline. In the follow-up surveys, mean prevalence was 37.7%, mean incidence 19.6%, and mean recurrence 69.0%. The most frequently observed courses across 5 years were those with a constant status: BP always absent (n = 1346, 34.7%) or BP always present (n = 538, 13.9%). BP recurrences increased with increasing numbers of previous consecutive years with BP from 46.9% (1 year of previous BP) to 88.1% (at least 4 years of previous BP). CONCLUSION.: BP history is highly predictive for future BP episodes.