Peripheral neuropathy has been recognized as a dose-limiting adverse effect in Phase I studies of didanosine (ddI) therapy for HIV infection. To study the effect of the currently recommended lower dose of ddI, the databases of 4 randomized, controlled trials were used to assess the frequency of dose-limiting peripheral neuropathy during treatment with ddI 500 or 750 mg/d, compared with zidovudine (ZDV) monotherapy or combination therapy with ddI/ZDV or zalcitabine/ZDV. No between-group differences in risk factors for neuropathy (eg, infectious and metabolic factors, malignancy, concurrent medications) were observed in the individual trials, and the presence of these risk factors appeared to have no increased treatment effect on the occurrence of neuropathy. No significant between-group differences were observed in the individual studies with regard to the incidence or time to onset of peripheral neuropathy. Analysis of the combined results by treatment regimen showed no significant difference in the incidence of neuropathy between recipients of ddI 500 mg/d, ddI 750 mg/d, or ZDV and no significant difference in the cumulative dose received until the onset of neuropathy between the ddI 500- and 750-mg regimens. Entry CD4+ cell counts were significantly predictive of neuropathy, with each 100-cell/microL decrement associated with a 17% increase in risk (P = 0.002); a CD4+ cell count of <50 cells/microL was highly predictive of neuropathy (P = 0.0001). In summary, the risk for peripheral neuropathy was not increased by treatment with ddI versus comparator regimens or by treatment with ddI at the dosages used in studies conducted more recently than the Phase I trials. Peripheral neuropathy seems more likely to be associated with advanced HIV infection and lower CD4+ cell counts (particularly counts <50 cells/microL) than with ddI therapy at the currently recommended dose.