Simulated throat swabs were prepared with known numbers of Streptococcus pyogenes; some were suspended in pasteurised human saliva and an equal number in saline. Two types of commercially available swabs were tested; these were composed of (1) plain, buffered, cotton wool, and (2) albumen-coated cotton wool. The mean recovery rates of first platings on solid media from albumen-coated and plain cotton-wool swabs were similar (8-2% and 8-3%) and the mean recovery rates from platings 1-4 were also similar (6-6% and 6-5%). The greater the delay in plating, the less were the chances of recovery of streptococci, although the viability of these was significantly prolonged on swabs held at 4 degrees C. Similar results were again produced by both types of swabs; processing swabs in saliva, however, produced by both types of swabs; processing swabs in saliva, however, produced a recovery rate that was 1-2% greater than the rate for the saline series. Swabs were also agitated or squeezed by forceps in nutrient broth to release any organisms they contained, and standard samples of the broth were then plated on solid media. Counts thus obtained indicated that about 50% of the original inoculum was still viable and could be recovered from the broth; in comparison, recoveries from initial direct plating of the swabs were low. No demonstrable toxic effect was produced by the cotton wool in these experiments.