As now constituted, the human tumor cloning assay performed in Petri dishes has several limitations including: (a) not all patients' tumors form colonies in the assay; (b) the plating efficiencies (number of colonies formed/number of cells plated) are low; and (c) a large number of tumor cells are required to perform drug sensitivity testing. In this study the use of capillary tubes, as vessels in which to clone human tumors, is compared to the use of 35-mm Petri dishes. In 100-microliters capillary tubes the optimal plating efficiencies are found with 50,000 cells/vessel (500,000 cells/ml), while in 35-mm Petri dishes the optimal plating efficiencies are found with 500,000 cells/vessel (250,000 cells/ml). In head to head comparisons of plating efficiencies of 183 human tumors (18 different histological types), the median plating efficiency was 5-fold higher (range, 1.16-37.00) for the capillary tubes than for the Petri dishes. This improved plating efficiency was noted for nearly all of the histological tumor types examined. The improved plating efficiencies noted with the capillary system indicate that the Petri dish method may be too selective and not reflect the total number of clonogenic units in a human tumor. In addition, the higher plating efficiencies noted with the capillary system may be exploited to solve some of the problems noted with the conventional Petri dish method.