This study aimed to determine whether sodium valproate (VPA) improves cognitive performance and behaviour in children with learning and behavioural problems associated with electrographic epileptiform discharges but without clinical seizures. A randomized, double-blind, single-crossover trial was carried out with VPA or placebo on eight participants with different learning and behaviour problems. Participants also underwent neuropsychological testing under video EEG and the parent and teacher Behaviour Check List (CBCL; Achenbach 1991a, b) during each treatment phase. Clinically none of the children improved on VPA. On formal testing children were more distractable, had increased delay in response time, and showed lower memory scores while on VPA. In addition, parents reported higher internalizing scores on the CBCL while children were on VPA. Our data do not support the use of VPA in similar patients.