BACKGROUND Nausea and vomiting are preventable side effects of cancer chemotherapy for children. Antiemetics are essential, especially as treatment becomes more intensive. Many drugs are available, but adequate evidence-based recommendations are lacking. We aimed (1) to consider an evidence-based approach for pharmacological prophylaxis of chemotherapy-induced nausea and vomiting (CINV) in children, and (2) to compare this approach with antiemetic prescribing in two paediatric oncology centres. PROCEDURE Relevant publications (Medline, Embase, CancerLit:1966-2002) were critically evaluated using pre-defined criteria. Evidence-based statements summarising their findings were formulated, and evidence basis proposed. Current prescribing practice was then compared with this evidence basis in Welsh children under 16 receiving chemotherapy at Llandough Hospital, Cardiff or Alder Hey Children's Hospital, Liverpool between 1 January 2001 and 31 December 2001. RESULTS Of 213 studies retrieved, 82 provided evidence. Our evidence basis recommends combination 5HT3-antagonist/corticosteroid for highly emetogenic chemotherapy, 5HT3-antagonist alone for moderate emetogenicity, and no antiemetic for other chemotherapy. Forty-four children in Cardiff (0.6-16.9 yrs) and 14 in Liverpool (0.8-16.2 yrs) were included in the audit. Differences in prescribing practice between the centres were not significant. In 109/159 (69%) of chemotherapy courses (35, 87 and 100% of high, moderate and low emetogenicity, respectively), antiemetics were selected in accordance with evidence basis. Seventy percent of prescribed doses were as evidence basis recommended. CONCLUSIONS We present an evidence basis for prescribing prophylactic antiemetics to children undergoing chemotherapy. Prescribing practices in these two centres treating Welsh children were similar. Both differed from the evidence basis we propose. Deviations were greatest for regimens of high emetogenicity, where effective emetic control is most crucial.