Interlaboratory variability of valproic acid determinations.

19 pooled plasma specimens were sent as unknowns to 13 participating research laboratories. The interlaboratory variability between the results was very high. Only 4 out of 13 laboratories had 6--12% of their results within the 95% confidence limit for each sample. The precision of repeated determinations was fairly good. 6 out of 10 participating laboratories had a coefficient of variation of less then 5%. The reproducibility and the agreement between the different procedures for quantitative analysis of dipropylacetate is similar to that reported for other major antiepileptic drugs.