OBJECTIVE To evaluate serum beta-2 microglobulin (beta-2M) and other prognostic indicators in previously untreated low-grade lymphoma. METHODS Cohort study of 80 patients with uniformly treated low-grade lymphoma, followed for a median of 21 months. These 80 patients, all of whom had serum beta-2M drawn within 2 weeks before starting therapy, were derived from a cohort of 119 previously untreated patients entered into one of three clinical trials. METHODS Tertiary referral cancer center. METHODS Eighty previously untreated stage I to IV patients (mean age, 55 years). METHODS Treatment was given according to Ann Arbor stage: Patients in stage IV were treated with CHOP-bleomycin and maintained on interferon therapy; those in stage III received CHOP-bleomycin and radiotherapy; and those in stages I and II received COP-bleomycin and radiotherapy. METHODS Outcome was determined by assessing complete remission rate and time to treatment failure. Univariate and multivariate analyses were used. RESULTS The complete remission rate for patients with a beta-2M level of 3.0 mg/L or greater was 36% compared with 71% for those with a level of less than 3.0 mg/L. Using multivariate analysis that tested beta-2M as a continuous variable, it was selected as the most significant factor for complete response. The adjusted odds ratio was 0.285 (95% CI, 0.101 to 0.809). The Ann Arbor stage had marginal significance (adjusted odds ratio, 0.435; CI, 0.150 to 1.263). For time to treatment failure, beta-2M was the only variable retained in the multivariate model. At 42 months, no patient with a beta-2M level of 3.0 mg/L or greater was projected to be in remission as compared with 85% of patients with a beta-2M level of less than 3.0 mg/L. CONCLUSIONS The serum beta-2M level is a good predictor of complete response and time to treatment failure. A larger number of patients should be studied to clarify the role of other potentially independent variables such as stage and age.