The genesis of human follicular lymphoma (FL) is a multistep process. The initial event is thought to be the chromosomal translocation t(14;18)(q32;q21) juxtaposing the bcl-2 proto-oncogene with the immunoglobulin (Ig) H chain locus joining segment (JH) as an error of D-J or V-D joining in the pre-B cell. However, FL is recognized clinically as a tumor of surface Ig (sIg)-positive B cells with morphologic and phenotypic similarities to the centrocyte of the secondary immune response. Thus, additional steps must be involved in the clonal expansion of the FL tumor cell beyond the activation of bcl-2 as a consequence of the t(14;18) translocation. Like the normal centrocyte, somatic mutations accumulate in the variable (V) genes of FL tumor B cells. To determine if clonal expansion of FL occurs before or after the development of the malignant follicle, we sought to examine the evolution of the FL V gene from its unmutated germline (GL) counterpart. To obtain the GL gene we first cloned the productively rearranged V gene of patient MT FL and obtained the clone rMTF. A hybridization probe derived from the 2.1-kb region upstream of the V gene in clone rMTF identified a single band in Southern blot hybridization of GL DNA. This probe was used to screen a size-selected library, and candidate GL V genes were isolated. Two identical clones, MTGL1 and 2, proved to have upstream regions (USRs) that were colinear with the USR of the rMTF. Thus, the MTGL clones represent the unmutated GL V genes, which were productively rearranged in the MT FL. Comparison of the GL V gene sequence to a consensus of MT FL V gene sequences revealed 42 mutations, demonstrating that malignant clonal expansion occurred subsequent to the activation of somatic mutation, presumably in the malignant follicle. Furthermore, the individual FL V gene sequences segregated into two distinct patterns of mutation. The major population represented 71% of the clones, and the minor population 29%. To investigate possible mechanisms for the parallel selection of distinct tumor cell populations, we analyzed the pattern of silent and replacement mutations within the V gene sequences. We found that in the framework regions (FRs) of both populations there were significantly fewer replacement changes than expected, suggesting that negative selective pressure was maintaining the structural integrity of the sIg. In contrast, the complementarity determining regions (CDRs), which make up the antigen binding domain of Ig, had an excess of replacement changes, suggesting positive selection for altered ligand binding.