Previous studies concerning the usefulness of pleural fluid glucose levels in differentiating causes of pleural effusions have been conflicting. Gelenger and Wiggers (1949), Calnan et al(1951) and Barber et al(1957) concluded that the lower the level of pleural fluid glucose, the more likely was tuberculosis, and that tuberculosis was unlikely if the pleural fluid glucose level was more than 80 mg/100 ml. Light and Ball(1973), however, reported that in the great majority of tuberculous pleural fluids the glucose concentration was high rather than low, concluded that the pleural fluid glucose levels were not useful in the differential diagnosis of pleural effusion. In this study, pleural fluid glucose was determined in 46 pleural effusions from various causes to evaluate the usefulness in the differential diagnosis of pleural effusion. In addition, the protein concentration and the electrophoretic patterns of protein and amylases in pleural fluid was compared with that of serum. And the results were as follows. 1. The mean glucose concentration of pleural fluid was 80.8 mg/100 ml in 22 tuberculous origin, 92.5 mg/100 ml in 12 cancer patient and 70.4 mg/100 ml in 10 undiagnosed cases. In 2 cases of paragonimiasis the pleural fliud glucose levels were low (mean, 32.0 mg/100 ml). The percentage of pleural fluid protein to serum is about 75% in all disease groups and the protein level of tuberculous pleural fluid was significantly correlated with that of serum. 2. The disc eletrophoretic patterns of pleural fluid were almost similar with that of serum in all disease groups but the prealbumin fraction was not observed in pleural fluid. 3. With the isoelectric focusing, 4 to 7 isoamylase was observed in serum and the isoelectric point was ranged from pH 5.8 to 7.8 and isoelectic point of main fracticn is pH 7.2. The isoelectic focusing patterns of amylase of pleural fluid were identical to that of serum in all disease group. With the above results it is concluded that the pleural fluid is exudate of serum and that the glucose levels of pleural fluid are not useful in the differential diagnosis of pieural effusions.