A variety of pre-clinical and clinical data point toward high drug levels of retinoids being required to achieve optimal efficacy against neuroblastoma. The results of the Kohler trial reported in this issue demonstrate that low-dose 13-cis-RA does not have clinical efficacy against neuroblastoma in a setting of minimal residual disease. A comparison of the Kohler trial with the US CCG trial provides clinical evidence that high-dose levels of retinoids are optimal for treating minimal residual disease in neuroblastoma. The comparison of high-dose and low-dose 13-cis-RA studies in neuroblastoma suggests the intriguing possibility that high dose, pulse schedules of other retinoids could be effective as therapeutic and chemopreventive agents in diseases where low-dose, chronic retinoid administration was not effective. Pre-clinical and perhaps clinical studies of the latter concept should be considered. © 2000 Cancer Research Campaign