Wilson and coworkers (Wilson SB, Kent SC, Patton KT, Orban T, Jackson RA, Exley M, Porcelli S, Schatz DA, Atkinson MA, Balk SP, Strominger JL, Hafler DA: Extreme Th1 bias of invariant V alpha24J alpha Q T-cells in type 1 diabetes. Nature 391:177-181, 1998) have recently reported raised serum levels of interleukin-4 (IL-4) in anti-islet autoantibody-positive first-degree relatives of patients with type 1A diabetes who did not progress to diabetes. Protection from diabetes has been noted for several human lymphocyte antigen (HLA) alleles, such as HLA DR2-DQA1*0102-DQB1*0602. We, therefore, wanted to determine whether this cytokine phenotype was associated with HLA genes protective for type 1A diabetes. We used a two-site fluoroimmunoassay with the same monoclonal antibodies as those reported by Wilson et al. Using this assay, we have found evidence for human heterophile antibodies mimicking serum IL-4: all serum IL-4 reactivity was lost if mouse serum or mouse immunoglobulin were added to the assay; serum IL-4 activity was bound and then eluted by protein A/G chromatography; and levels of anti-mouse antibodies correlated with apparent serum IL-4. This pseudo-IL-4 activity was found in a subset of control subjects, patients with type 1A diabetes, and their relatives and was primarily associated with specific HLA alleles protective for type 1A diabetes (e.g., DQB1*0602). After adjustment for HLA, positive levels of heterophile antibodies were not associated with protection from diabetes. The confounding effect of protective HLA alleles associated with heterophile antibodies could explain the previously reported association between raised serum IL-4 and protection from type 1A diabetes. The mechanism by which specific DQ alleles protect from diabetes and are associated with increased heterophile antibodies is currently unknown.