To date, the clinical success of 13-cis or all-trans retinoic acid in the treatment of neuroblastoma has been disappointing. In vivo, 13-cis will isomerise to both all-trans and 9-cis retinoic acid, believed to be the main biologically-active isomers. In vitro studies with an N-type neuroblastoma cell line, SH SY 5Y, show that 9-cis is better than other isomers at both inducing morphological differentiation and inhibiting proliferation. RAR-beta, a gene which may mediate retinoic acid responsiveness and be of prognostic significance, is also more-effectively induced by 9-cis retinoic acid. 9-cis and all-trans retinoic acid do not have synergistic effects on SH SY 5Y cell proliferation and gene expression. A retinoid X receptor (RXR)-specific analogue of 9-cis retinoic acid had similar effects on gene expression to 9-cis retinoic acid alone. In view of these results, 9-cis retinoic acid or stable analogues of this retinoid may have potential for the treatment of neuroblastoma.