A prognostic factor analysis for survival and response was conducted on 374 previously treated patients who were treated with one of four fludarabine regimens. As several prognostic factors were associated with response and survival, multivariate analyses were conducted to identify the combination of factors which best describe the prognosis as regards response to fludarabine and survival. Statistical models to identify the hazard for survival and probability of response were developed. The characteristics which were included in the model predicting for survival were sex, age, number of prior treatments, performance status, hemoglobin level, serum albumin, and alkaline phosphatase levels. The characteristics most strongly associated with response were the hemoglobin levels, serum albumin, and the number of prior treatments. These models can be used to identify risk groups to guide physicians in decision-making and to assist in the design and analysis of clinical trials.