We compared the sensitivity and specificity of the phenol ammonium sulfate (PhAS) sediment smear microscopy method for detection of acid-fast bacilli with those of direct smear microscopy, using culture results for Mycobacterium tuberculosis as the "gold standard." The sensitivities of the PhAS and direct smear methods were 85% (465 of 547) and 83% (454 of 547), respectively, and the specificity of each method was 97%. The PhAS method was better accepted by the laboratory technicians and safer but necessitates an overnight sedimentation, which delays reporting of results until 1 day after sputum collection.Each year, there are an estimated 2 million tuberculosis deaths and 8 million new cases of tuberculosis worldwide (3, 13). In developing countries, diagnosis of pulmonary tuberculosis depends primarily on the identification of acid-fast bacilli (AFB) using Ziehl-Neelsen sputum smear microscopy, a technique more than 100 years old. The sensitivity of this method varies (7) and depends upon collection of sufficient sputum, proper preparation of smears, good staining technique, careful examination of smears, and availability of a good microscope. Several methods have been tried to improve smear microscopy for AFB (1, 4-6, 9, 11, 12), but these have limitations under field conditions. Developing new smear microscopy methods which would be feasible under field conditions remains a priority for improving the diagnosis of pulmonary tuberculosis. In developing countries, laboratory technicians sometimes tend to sidestep the sputum examination owing to apprehensions about the infectiousness of sputum samples and due to the cumbersome method of preparing direct smears from the mucus portion of the sample. We evaluated the sensitivity, specificity, and acceptability of a new smear microscopy method using phenol ammonium sulfate (PhAS) for sedimentation of sputum.The study was conducted at a large tertiary-care hospital in Chennai, Tamil Nadu, South India. Each patient attending the outpatient department of the hospital was screened for chest symptoms. One spot sputum specimen was collected in a McCartney bottle from each patient who reported having a cough for 3 weeks or longer. Patients who were on antituberculosis treatment at the time were excluded from the study. All samples were collected during the months of August and September 2001.Processing of sputum samples. Sputum specimens were transported and processed within 4 h of collection. Each sample was divided into two portions of 3 to 5 ml each by pouring the sample from one McCartney bottle to another, such that both portions were approximately equal in volume and appeared similar in quality. The two portions were randomly allocated; one to the modified Petroff method for culture of Mycobacterium tuberculosis (10) and the other to the PhAS sediment smear method. For the Petroff method, sputum was homogenized for 15 min in a shaker by using an equal volume of 4% sodium hydroxide. After centrifugation at 3,000 rpm for 15 min in a Megafuge 1.0 (Heraeus), the deposit ...