To investigate the natural history of stage A chronic lymphocytic leukemia (CLL) we reviewed 84 such patients. Among 74 cases evaluable for disease progression, 22 (29.6%) progressed to more advanced clinical stages (9 to stage B, 13 to stage C); the actuarial estimation of such an event at 4 years was 30% (95% CI: 26.3% to 33.6%). Despite a linear trend toward an increasing risk (r = .92), the hazard function analysis showed a constant pattern of progression, suggesting a lack of correlation of such an event with time (r = .04). Furthermore, disease progression when analyzed as a time-dependent variable had a clear-cut impact on survival (P less than .001). With the aim of identifying a subgroup of patients with low probability of disease progression and death, we applied to our set of patients four different proposals for subclassifying stage A. All methods were similar in terms of sample size, 5-year survival rate, and disease progression risk, suggesting that the choice between different proposals is somewhat arbitrary. Whatever the criteria are for defining "smoldering" CLL, such patients (accounting in the present study for 20.5% of overall series and 46.7% of stage A patients) should not be treated until progression occurs.