Purpose: To evaluate how referrals from Norwegian general practitioners for plain radiography of the lumbosacral spine conform to clinical recommendations, and whether non-conforming referrals yield important findings. Material and Methods: The clinical information reported in referral letters of 323 patients was compared with recommendations given in a Norwegian and in a British “guide”. Additional clinical data were obtained by interviewing 100 of the patients. Using all available information, a new comparison was performed. Radiological findings were obtained from the routine reports. Results: of the 323 referrals, 24% (37%) conformed to the Norwegian (British) “guide”, 34% (46%) did not conform, and 42% (18%) were considered uncertain, mainly because of lack of pertinent information in the referral letters. A total of 182 examinations were in disagreement with one or both “guides”. Only 4 of these examinations revealed potentially important findings (osteoporotic fractures in 2 patients, uncertain sacroiliac joint arthritis in 1 patient, and “probably benign” sclerotic densities in 1 patient). In the interview group, the proportion of non-conforming referrals was 40% (48%) based on the referral letters, and 31% (30%) when the interview data were taken into account. Conclusion: A great proportion of referrals for plain radiography of the lumbosacral spine do not conform to recent clinical recommendations. Referrals outside the advised criteria yield few relevant findings and could probably be cancelled. They require at least supplementary information to justify radiography.