In order to assess whether DNA rearrangement occurs with equal frequency at each of the several J (joining region) elements in the mouse kappa light chain locus, we set out to determine the relative frequency of usage of J kappa segments in populations of B lymphocytes unperturbed by antigenic selection or cloning. To obtain such a population, we exposed a suspension of spleen cells to a mixture of mitogens capable of activating most B cells independently of their specificity for antigen. We estimated the relative usage of the J kappa elements in unspliced kappa chain gene transcripts in total and poly(A)-containing nuclear RNA, using an S1 nuclease protection assay, and in mature kappa chain mRNA, using a specifically primed cDNA hybridization assay. Both types of assay reveal a marked difference in the frequency of J kappa elements and indicate that their relative usage is: J1 approximately J2 much greater than J4 approximately J3. Comparison of the 5' flanking regions of the mouse J kappa elements, including the conserved putative recombination target sequences, shows no obvious differences consistent with the variation in recombinational efficiency, so we conclude that, although the consensus heptamer and nonamer signals may be sufficient to identify a recombination site, the probability that that site will be used depends also on other determinants. A review of published data suggests a nonequivalence of usage also among human J kappa elements and between mouse J lambda I and J lambda III loci. Extending the comparison to include these sets of sequences indicates that one of the determinants of frequency of J use may be the proximity of the consensus heptamer to a T-G dinucleotide within the J coding sequence, perhaps revealing an Escherichia coli gyrase-like substrate preference within a recombination enzyme.