Most patients referred to physiotherapy with low back pain are without a precise medical diagnosis. Identification of subgroups of non-specific low back pain patients may improve clinical outcomes and research efficiency. A pathoanatomic classification system has been developed, classifying patients with non-specific low back pain into 12 different syndromes and three subcategories based on history and physical examination. The purpose of this study was to estimate the inter-tester reliability of clinical tests used as criteria for classifying patients. Ninety patients with chronic low back pain were each examined by two physiotherapists. A total of four physiotherapists conducted the assessments. Examination findings were recorded independently by the two examiners. Percentage of agreement and kappa coefficients were calculated for each category. The overall rate of agreement was 72% and the kappa coefficient was 0.62 for the mutually exclusive syndromes in the classification system. Agreement rates for each of the syndromes ranged from 74% to 100% and kappa coefficients ranged from 0.44 to 1.00. The findings suggest the inter-tester reliability of the system is acceptable. The relatively modest level of total agreement (39%) for the system as a whole might indicate that the utility of the system for general screening purposes is limited, compared with the utility in identification of particular syndromes. Due to low prevalence of positive findings in some of the syndromes, future work should focus on testing reliability on a larger sample of patients, and testing of validity and feasibility of the system.