The aim of the study was to investigate the effect of carnitine supplementation of valproic acid (VPA) treated patients presenting with increased plasma ammonia concentrations. Plasma ammonia concentrations were recorded in 69 children and young adults on VPA monotherapy (25.6 +/- 9.2 mg VPA/kg per day; mean plasma VPA concentration 68.8 +/- 27.6 mg/l). Their mean plasma ammonia concentration was 80.2 +/- 32.1 micrograms/dl (median 73.1 microgram/dl). A total of 24 patients (35.3%) presenting with ammonia concentrations > 80 microgram/dl were considered hyperammonaemic. Of these, 15/24 (22.1%) showed ammonia concentrations > 100 microgram/dl, even up to 194 micrograms/dl. In 48/69 patients, plasma carnitine concentrations could be determined. The plasma total carnitine (TC) concentrations were rather low (26.9 +/- 8.8 mumol/1) compared to normal values obtained in our laboratory (40.9 +/- 7.2 mumol/1). The percentage of free carnitine was considered decreased (< 75% TC) in 13/48 samples (27%). Fourteen hyperammonaemic patients and one with a plasma ammonia level of 60 micrograms/dl agreed to be supplemented with L-carnitine (1 g/m2 per day divided into two equal doses). Their plasma ammonia and carnitine concentrations were re-evaluated after a mean of 9.1 +/- 4.0 days (median 9.0 days) and in 9 patients again after a mean of 79.6 +/- 30.1 days (median 75 days) of L-carnitine supplementation. Plasma ammonia concentrations decreased in all 15 patients. The decrease was 25.4 +/- 11.2% (median 28.3%) after a mean of 9.1 +/- 4.0 days and amounted to 46.0 +/- 17.2% (median 48%) after 79.6 +/- 30.1 days. L-Carnitine supplementation led to an increase in plasma free carnitine of 11.6 +/- 13.0% (median 15.6%) and to a further increase of 11.1 +/- 8.4% (median 11.5%) when re-evaluated a second time. The plasma ammonia concentrations were significantly correlated with the percentage of free plasma carnitine (r = -0.67; p < 0.0001). The results show that carnitine supplementation is a means of normalizing elevated plasma ammonia concentrations. However, we cannot conclude from our results whether this lowers the risk of developing a VPA-induced Reye's-like syndrome.