OBJECTIVE: To determine whether measuring monoclonal bands (M component) in serum should be part of the investigation of patients referred to osteoporosis clinics. DESIGN: Retrospective, cross sectional, observational study. SETTING: Referral centre for osteoporosis in a university hospital, Denmark. PARTICIPANTS: 799 people (685 women) aged 19 to 94 years newly referred with suspected osteoporosis. MAIN OUTCOME MEASURES: Proportion of patients fulfilling the Nordic Myeloma Study Group definition for target condition and proportion of patients with other important haematological conditions. RESULTS: 4.9% (18 of 366) of patients with osteoporosis and 2.2% (9 of 408) of patients without osteoporosis had M component in serum (chi2 = 3.66, P = 0.04). Multiple myeloma was diagnosed in three patients with osteoporosis (absolute risk 0.8%, 95% confidence interval 0.11% to 1.7%). The relative risk of multiple myeloma in patients presenting with osteoporosis was 75 (10 to 160). As a diagnostic test for multiple myeloma in patients with osteoporosis, M component in serum had a specificity of 95.0% and a positive predictive value of 17.6%. 122 blood electrophoreses were carried out for each case of multiple myeloma diagnosed. All patients with multiple myeloma had a history of fragility fractures. If lymphoma was included as a target condition, the specificity increased to 95.3% and the positive predictive value increased to 23.5%. Monoclonal gammopathy of undetermined significance was diagnosed in 13 (3.6%) participants with osteoporosis and in eight (2.0%) participants with normal bone mineral density or osteopenia. CONCLUSIONS: Patients presenting with osteoporosis should be tested for M component in serum, as 1 in 20 patients with newly diagnosed osteoporosis had multiple myeloma or monoclonal gammopathy of undetermined significance.