Data from the three preceding studies on the possible relationship of HL-A type with Burkitt's lymphoma were combined to see whether any trends in the data, not significant in the individal studies, would show significance when the three groups were combined. A statistical method for combining data from different studies is developed. Although in this analysis even the combined data failed to show any significant association of HL-A type with disease, it is suggested that the statistical methods used will be helpful in analysing other sets of data where it seems desirable to combine information from several studies