Mutations that cause X-linked agammaglobulinemia (XLA) appear throughout the Bruton tyrosine kinase (Btk) sequence, including the pleckstrin homology (PH) domain. To analyze the basis of this disease with respect to protein structure, we studied the relationships between PH domain sequences and structures by comparing sequence-based profiles of physicochemical properties and solvent accessibility profiles. The diversity of the distribution of amino acids was measured by calculating entropies for sequences containing mutations at different positions in multiple sequence alignments. Mutual information was calculated to quantify positional covariation. Eight conserved extrema were apparent in all profiles. The majority of the XLA disease-causing mutations in the Btk PH domain were found at positions having significant mutual information, indicating that there are covariant constraints for both structure and function. Together with additional structural analyses, all the XLA mutations that were analyzed could be explained at the molecular level. The method developed here is applicable to the design of mutations for protein engineering.