The feasibility of using an image analyser, the CAS 100, to assess the ploidy of smooth muscle tumours on formalin fixed paraffin wax embedded tissue was assessed. Various different methods of assessment were compared. A paraffin wax block of 15 leiomyomata and 13 leiomyosarcomata yielded serial sections of 3 microns, 5 microns, and 7 microns and a cytospin preparation (from a 50 micron section). Sections and cytospin preparations were stained with Feulgen and quantified using the CAS 100. The suitability of lymphocytes, previously suggested to be unsuitable as control cells in tissue sections due to the compact nature of their DNA, was assessed in sections and cytospin preparations and compared with endothelial cells, the standard alternative, on the same slide. Despite having a mean nuclear area of only 18.5 microns 2 in sections lymphocytes had a similar diploid peak--that is, 4.1 pg--to endothelial cells (mean nuclear area 39 microns 2). A comparison of 3 microns, 5 microns, and 7 microns sections showed 5 microns to be the optimal thickness. Cytospin preparations yielded histograms of superior quality than those from tissue sections. All 15 leiomyomata had a 5C exceeding rate of less than 0.3%. Ten of 11 histologically malignant tumours had a 5C exceeding rate of more than 5% (mean 14%) in sections and cytospins. Two leiomyosarcomata that had arisen in the lower gastrointestinal tract had a mitotic rate of less than or equal to 1/10 per high power fields and yielded histograms similar to those of the leiomyomata. It is concluded that formalin fixed, paraffin wax embedded tissue can be used for DNA quantification by image analysis; that tissue sections yield poorer results than cytospin preparations; that lymphocytes are reliable control cells in cytospin preparations; and that a population of cell greater than 5C is seen in 90% of leiomyosarcomata.