Our objective was to evaluate the effect of intravenous immunoglobulin (IVIG) on absolute CD4+ lymphocyte count (CD4+ count) trends in human immunodeficiency virus- (HIV) infected children enrolled in a trial of IVIG for infection prophylaxis. To that end, we conducted a randomized, double-blind, outpatient trial comparing subjects treated with 400 mg per kilogram of IVIG every 28 days with those given 0.1% albumin placebo. CD4+ counts were measured at entry and every 12 weeks. Twenty-eight clinical centers in mainland United States and Puerto Rico participated. Previous reports showed IVIG efficacy for infection prophylaxis in 313 patients with entry CD4+ counts of > or = 0.20 x 10(9)/L (> or = 200/mm3). Two hundred and seventy-seven (89%) of these 313 children had three or more CD4+ counts measured during the trial and were included in evaluation of CD4+ count trends. Rates of CD4+ count decline, as measured by regression slopes, were compared between IVIG and placebo groups using generalized linear models, comparing unadjusted, age-adjusted, and standardized age-adjusted data. Potential covariate effects were assessed by modeling change in CD4+ count in terms of log change between successive measurements. Age-adjusted slope analysis showed slowing of CD4+ count decline by 13.5 cells/mm3 per month in IVIG compared with placebo recipients (95% confidence interval, 3.1-23.9, p = 0.012). Modeling log change between measurements documented a beneficial effect of IVIG that was cumulative over time and independent of other therapies. Occurrence of serious bacterial infection in the interval before CD4+ count measurement or death was independently associated with more rapid CD4+ count decline (p = 0.01 and p = 0.008, respectively). Zidovudine therapy was associated with a transient increase in CD4+ count. Benefits of IVIG include slowing of CD4+ count decline as well as previously reported reductions in serious and minor bacterial and viral infections in subjects with entry CD4+ counts of > or = 0.20 x 10(9)/L. This finding provides corroboration for the hypothesis that immunologic mechanisms contribute to the pathogenesis of CD4+ decline in HIV infection.