Previous work has shown platinum drugs to differ in their effects on the peripheral nervous system. To test whether their differential toxicity was due to differences in their partitioning into the peripheral nervous system, we correlated the hydrophobicity, reactivity, tissue accumulation and neurotoxicity of a series of eight platinum analogues. Neurotoxicity was detected by measuring sensory nerve conduction velocity (SNCV) in Wistar rats treated twice per week at the maximum tolerated dose. Tissue platinum concentrations were measured by inductively coupled plasma mass spectrometry. Hydrophobicity (log P) was measured using an octanol-aqueous shake-flask method. The half-life of platinum drug binding to plasma proteins in vitro was determined. The cumulative dose causing altered SNCV ranged from 15 to > 2050 μmol kg−1. Ranking of the compounds by their neurotoxic potency in rats (oxaliplatin >R,R -(DACH)PtC4> ormaplatin >S,S -(DACH)PtCl4>S,S -(DACH)Pt oxalato > cisplatin > carboplatin > JM216) correlated with the frequency of neurotoxicity in patients (r> 0.99;P< 0.05). Ranking the compounds by their peripheral nerve accumulation was cisplatin > carboplatin > oxaliplatin >R,R -(DACH)PtCl4≈S,S -(DACH)PtCl4and did not correlate with neurotoxicity. Log P ranged from – 2.53 to –0.16 but did not correlate with neurotoxicity. Log P correlated inversely with platinum accumulation in dorsal root ganglia (r2= 0.99;P = 0.04), sural nerve (r2= 0.85;P = 0.025), sciatic nerve (r2= 0.98;P = 0.0012), spinal cord (r2= 0.97, P = 0.018) and brain (r2= 0.98, P = 0.001). Reactivity correlated with neurotoxicity potency in rats (r2= 0.89, P = 0.0005) and with the frequency of neurotoxicity in patients (r2= 0.99, P = 0.0002). The hydrophilicity of platinum drugs correlates with platinum sequestration in the peripheral nervous system but not with neurotoxicity. Differences in the reactivity of platinum complexes accounts for some of the variation in their neurotoxicity. © 2000 Cancer Research Campaign