To the Editor: The letter of Kornfeld et al. raises two separate questions: Do some patients with mild hypogammaglobulinemia have defects in the same gene as patients with typical X-linked agammaglobulinemia? Can you predict the clinical phenotype of X-linked agammaglobulinemia on the basis of the site of the Btk mutation? Our paper answers the first question by demonstrating that a mutation in Btk can cause an atypical or mild form of X-linked agammaglobulinemia. The results described by Kornfeld et al. support this conclusion. We did not address the second question; however, emerging data indicate that the genotype-phenotype correlations in X-linked agammaglobulinemia are likely to be complex. Within a single family, such as our patient's family, some members may have mild X-linked agammaglobulinemia whereas others have more typical disease, indicating that factors other than the site of mutation influence the phenotype. These factors may include past infections, the age of the patient, and the treatment received, as well as modifying or compensatory genetic factors. As suggested by Kornfeld et al., mutations in non-kinase domains of Btk have been associated with typical X-linked agammaglobulinemia and mutations in the kinase domain have been associated with mild disease1. However, we would argue that both the site and the type of Btk mutation make important contributions to the phenotype. We were surprised to find that a high percentage of our patients with mild disease lacked the Btk transcript in cell lines transformed by Epstein-Barr virus or neutrophils. To identify the mutations in these patients, we screened genomic DNA by single-strand conformation polymorphism using polymerase-chain-reaction primers that flanked each of the 19 exons of Btk2. The patients lacking the transcript were found to have premature stop codons as the result of either substitutions in a single base pair or small insertions or deletions associated with frame shifts. It is well recognized that premature stop codons in some genes, for example β-globin, are associated with a failure to accumulate transcript in the cytoplasm3. As was true with Kornfeld's patients, some but not all of our patients with premature stop codons had mild disease4. We interpret this finding as indicating that in the absence of Btk, other kinases may partially compensate. This leads to the further observation that some mutant Btk proteins may be more deleterious to B-cell proliferation or survival than a complete absence of Btk. As gene therapy and other sophisticated treatments for X-linked agammaglobulinemia are contemplated, it becomes increasingly important to understand how all the factors that contribute to the phenotype for this deficiency interact.