Ninety-nine anteroposterior and lateral lumbar radiographs taken of men, 18 to 55 years of age, were randomly selected from participants in a population survey of low-back pain. Thirty-one (31%) had never had low-back pain; 44 (44%) had or were having mild low-back pain; and 24 (24%) had or were having severe low-back pain. Three chiropractors assessed 56 radiographic variables, which included determinations of disc space height, vertebral malalignments and subluxations, spondylosis, postural disturbances, relationships among pelvis and spine and other congenital or acquired abnormalities. Interobserver reliability meaasurements showed that 6 of the 56 variables analyzed produced a high interobserver reliability. Sixteen additional variables showed a fair interobserver reliability. Comparison of the observations made by the chiropractors and a radiologist showed minimal agreement except for disc space height assessments at L3-4 and L4-5. Few of the radiographic variables discriminated between the current or prior history of low-back and leg complaints, although a few variables (most notably disc space narrowing) were statistically associated with back or leg complaints (P = .025). The conclusion was reached that spinal radiographs, whether analyzed by measurements, by a radiologist, or by chiropractors, have minimal value in determining the presence or absence of low-back complaints and, in particular, have no value in epidemiologic studies.