With the emergence of the human immunodeficiency virus (HIV) epidemic, lymphocyte immunophenotyping has become the single most important laboratory test for clinical management of HIV-infected subjects. To meet this challenge, the department of Army instituted a multicenter lymphocyte immunophenotyping quality assurance (QA) program in March 1986. An integral part of the QA program has been the development of a monthly proficiency testing program to survey the degree of precision and reproducibility of lymphocyte subset determinations within the Army. After 15 months of proficiency testing, the multicenter cumulative average standard deviation for the percentage of positive CD2 was 3.3, CD3 was 4.4, CD4 was 3.3, CD8 was 3.6, CD8*CD3 was 2.8, CD19/20 was 2.9, and 3.0 for natural killer (NK) cells. The cumulative average coefficient of variation for the percentage of positive CD2 was 3.9%, CD3 was 4.9%, CD4 was 6.6%, CD8 was 11.4%, CD8*CD3 was 9.4%, CD19/20 was 18.8%, and 26.5% for NK. Five survey shipments were also shipped to an additional 49 laboratories outside the Department of Army. The difference of the mean Army percentage positive values from the mean overall percentage positive values ranged from zero to 9.6, with an average difference of 1.6. The interlaboratory variability of flow cytometrically-derived percentage values presented in this report are almost half that cited by other multicenter lymphocyte comparative studies.