The choice of cytotoxic drugs given to patients with a certain type of cancer is usually based on previous experience gained in the treatment of patients with the same type of cancer. In the last two decades much effort has been directed to obtain a rational basis for the selection of anticancer drugs and to improve the results of cytotoxic chemotherapy.Based on an extensive review of the literature on in vivo drug sensitivity testing, sensitivity testing in animals was not considered appropriate for individual drug testing, although the behaviour of drugs in animals is highly comparable to the human situation. After reviewing the literature on in vitro sensitivity testing, we chose a human tumour colony-forming assay for further studies, as colony-forming assays selectively promote the growth of 'tumour stem cells'. This culture technique starts with single cells (malignant and non-malignant cells) in suspension, which are incubated with and without drugs for a certain period or during the complete culture period. The cells are plated in the upper layer of a bilayered Petri dish. Both layers contain agar and nutrients. After I4-2I days in a humidified incubator with 5-7.5% CO2, colonies may have grown. Colonies are defined as groups of thirty or more cells, which are regarded as having originated from a single malignant cell after five or more cell divisions.We started to study the possibilities of the human tumour colony-forming assay as described by Hamburger and Salmon. As plating efficiency (defined as percentage of colonies per plated cells) and adequate growth rate (defined as more than thirty colonies per dish) were low, possibilities to improve the culture results were investigated. The main change was the addition of batches of cell-free ascites fluid originating from cancer patients which had shown a stimulating effect on colony growth of several cell lines. This resulted in a higher plating efficiency and a higher adequate growth rate for fresh specimens in spite of a lower number of cells plated.This slightly modified technique was used to culture I4o fresh tumour specimens. The overall adequate growth rate was 40%; the adequate growth in the absence of cell-free ascites fluid 27% (I 2 out of 44 specimens) and in the presence of 25% cell-free ascites fluid 46% (42 out of 90). The mean plating efficiency was 0.09% (= 9 colonies per IO,OOO cells plated). Such a low plating efficiency suggests that the culture conditions may select for a fraction of stem cells. In that case the representative nature of the assay system is markedly reduced.Another culture problem was the difficulty in obtaining a viable single cell suspension. Clumps or aggregates are difficult to dissociate and single cells may reaggregate in the period around plating. For a drug sensitivity assay the number of cells should be proportional to the number of colonies. Most authors who report on this relationship assume it to be present.We cultured 28 fresh specimens at different concentrations of plated cells. For thirteen of these specim...