I-131-radiolabeled tositumomab (Anti-B1 Antibody), in conjunction with unlabeled tositumomab, was employed in a phase II clinical trial for the therapy of 76 previously-untreated follicular-non-Hodgkin's-lymphoma patients at the University of Michigan Cancer Center. For all patients, conjugate-view images were obtained at six to eight time points on seven consecutive days after a tracer infusion of the antibody. A SPECT image set was obtained on day two or three after the therapy infusion for 57 of the patients. Of these, 55 are suitable for dosimetric evaluation. To date, we have completed analysis and response characterization of 20 patients from the subset of 55. All 20 patients had either a complete response (CR) or a partial response (PR). Conjugate-views provided a time-activity curve for a composite of nearby, individual tumors. These tumors were unresolved in the anterior-posterior projection. Pre-therapy CT provided volume estimates. Therapy radiation dose was computed for the composite tumor by standard MIRD methods. Intra-therapy SPECT allowed the calculation of a separate dose estimate for each individual tumor associated with the composite tumor. Average dose estimates for each patient were also calculated. The 30 individual tumors in PR patients had a mean radiation dose of (369 +/- 54) cGy, while the 56 individual tumors in CR patients had a mean radiation dose of (720 +/- 80) cGy. According to a mixed ANOVA analysis, there was a trend toward a significant difference between the radiation dose absorbed by individual tumors for PR patients and that for CR patients. When the radiation dose depended on only the patient response, the p value was 0.04. When the radiation dose depended on the pre-therapy volume of the individual tumor as well as on the patient response, the p value was 0.06. Since the patient response was complete in 75% of the patients, the analysis of the total cohort of 55 evaluable patients is needed to have a larger number of PR patients to better test the trend toward a significant difference. A pseudo-prediction analysis for patient-level dose and response was also carried out. The positive predictive value and the negative predictive value were 73% and 80%, respectively when a patient's average radiation dose was used. The predictive values were 73% and 60%, respectively, when the patient's average base-10 logarithm of radiation dose was used. A complete overlap for the dose range of CR patients compared to that for PR patients precluded higher predictive values. In conclusion, there was a trend toward a significant difference in the radiation dose between CR and PR patients, but it was only moderately predictive of response.