Syntheses are described for and structure:activity studies undertaken of the anti-tumour activity of (2-crotonyloxymethyl-(4R,5R,6R)-4,5,6-trihydroxycyclohex-2-+ ++enone) (1) and its analogues 1-crotonyloxymethyl-(3R,4S,5R)-3,4,5-trihydroxycyclohex-1-en e (3), 1-crotonyloxymethyl-(3R,4S,5S)-3,4,5-trihydroxycyclohexene (4) and 2-crotonyloxymethyl-2-cyclohexenone (5), which differ from 1 in the presence/absence of the cyclic keto group and/or the stereochemistry at one of the -OH bearing carbon atoms. None of the above compounds, including 1, directly inhibited glyoxalase I, isolated for the first time to homogeneity from rat Yoshida sarcomas and for which a purification protocol was developed. The apparent inhibition of glyoxalase I by 1 and 5 (but not detected for 4 or 3) could be explained by reaction of 1 and 5 with the glutathione present in the assay buffer and the consequent depletion of substrate. 1 and 5 were found to react readily with glutathione whereas 4 and 3 did not react. In vitro chemosensitivity studies against a panel of tumour cell lines of both mouse and human origin showed that in parallel with their thiol reactivity, 1 and 5 exhibited significant in vitro cytotoxicity whereas 4 and 3 did not. Concentrations of drug required to cause 50% cell kill (ID50 values) were in the range 0.5-19 microM (0.1-2.8 micrograms/ml) for 5, and 3-44 microM (0.7-10 micrograms/ml) for 1. The structural features causing the differences in antitumour effects were localized on this basis to the alpha,beta-unsaturated ketone linkage as opposed to the stereochemistry of the (trihydroxy) alcoholic sites.