This study was designed to assess the potential benefit of hydroxyurea (HU) on virological and immunological markers in heavily pretreated human immunodeficiency virus (HIV)-infected patients. A total of 118 patients received hydroxyurea as treatment added to their current antiviral therapy ('add' group), or as treatment added together with a switched nucleoside reverse transcriptase inhibitor or protease inhibitor or with the introduction of an additional antiviral agent ('add and switch' group). Efficacy of HU was also compared among patients receiving didanosine- and non-didanosine-containing regimens. Results showed that HU was not associated with a significant decline in plasma HIV-1 RNA, and no significant difference in plasma HIV-1 RNA changes was noted between patients who did and did not receive didanosine. The absolute number of CD4 cells, but not the CD4 percentage, declined significantly from baseline. Significantly more grade 3-4 neutropenia occurred among patients receiving HU combined with zidovudine, although the discontinuation rate for haematological toxicity was similar in zidovudine- and non-zidovudine-treated patients.