BACKGROUND: Peripheral neuropathy is the dose-limiting toxicity of stavudine and didanosine (nucleoside analogs used in HIV treatment) and is attributed to mitochondrial toxicity from these drugs. Acetyl L-carnitine (ALC) and co-enzyme Q(10) are proposed as neuropathy treatments, but evidence to support these is limited. METHODS: We examined ALC and a water-soluble formulation of co-enzyme Q(10) (H(Q)O) for the prevention of d4T and ddI neurotoxicity using cultured fetal rat DRG as an in vitro model. RESULTS: DdI (33microM) and d4T (50microM) caused clear toxicity (impaired neurite growth) by day 8 of DRG culture. H(Q)O at concentrations 1-100microM completely prevented the toxicity of 33microM ddI in vitro and ALC at concentrations 1-100 microM substantially (but incompletely) prevented ddI toxicity in this model. In contrast, ALC was ineffective at all concentrations tested for preventing the toxicity of 50microM d4T. H(Q)O showed dose-dependent efficacy for preventing d4T toxicity. H(Q)O (1microM) partially prevented d4T toxicity while 10 and 100microM H(Q)O completely prevented d4T toxicity in this model. CONCLUSIONS: We find H(Q)O is superior to ALC for preventing the neurotoxicity of d4T (the HIV treatment most associated with neuropathy) and ddI in vitro. Further study is needed to clarify any clinical role for co-enzyme Q(10) co-administration with d4T and ddI and to assess whether this compound may have a role in treating established cases of neuropathy.