Cutaneous and uveal melanoma both have a poor prognosis and chemotherapy is usually unsuccessful. We have previously reported the activity of a number of cytotoxic agents against metastatic cutaneous and primary choroidal uveal melanoma using an ex vivo adenosine triphosphate (ATP)-based chemosensitivity assay (ATP-TCA). In this study we compare the results obtained with the two types of melanoma. Cutaneous melanoma deposits in skin and lymph nodes (n = 58) and choroidal melanomas (n = 77) were tested using the ATP-TCA. Analysis of the data based on an arbitrary threshold for sensitivity shows that both types of melanoma exhibit heterogeneity of sensitivity to all the agents and combinations tested. With all the single agents except gemcitabine, cutaneous melanomas showed greater sensitivity in the assay, though this did not achieve statistical significance. This was also true with the drug combinations, with the exception of treosulfan + gemcitabine, which had similar activity in each type of melanoma. Of all the single agents tested, doxorubicin (47% of specimens classed as sensitive), vinorelbine (43%), treosulfan (41%) and paclitaxel (33%) showed the greatest activity with cutaneous melanoma. In the uveal melanoma samples, mitoxantrone (33%), gemcitabine (22%) and treosulfan (21%) showed the greatest activity. In contrast to the cutaneous melanomas, 13% of the uveal melanomas were sensitive to paclitaxel, 4% were sensitive to doxorubicin and 11% were found to be sensitive to vinorelbine. Both tumour types showed greater sensitivity to combinations of cytotoxic agents. The combination of treosulfan + gemcitabine was universally effective, with 72% of cutaneous melanomas and 80% of uveal melanomas exhibiting activity at the level selected to indicate sensitivity in the assay, though this will not necessarily indicate a similar level of clinical sensitivity.