A systematic literature review of population prevalence studies of low back pain between 1966 and 1998 was conducted to investigate data homogeneity and appropriateness for pooling. Fifty-six studies were analyzed using methodologic criteria that examined sample representativeness, data quality, and pain definition. Acceptable studies were assessed for homogeneity and appropriateness for pooling. Thirty were methodologically acceptable. Of these there were significant differences in study design, patient age, mode of data collection, potential temporal effects, and prevalence results. Point prevalence ranged from 12% to 33%, 1-year prevalence ranged from 22% to 65%, and lifetime prevalence ranged from 11% to 84%. A limited number of studies were left for analysis, making the pooling of data difficult. A model using uniform best-practice methods is proposed.