Background: Transthoracic ultrasound (US) is an important instrument to identify pleural effusions and safely conduct invasive procedures. It also allows systematic scanning of the pleural surface, though its value remains uncertain for differentiation between malignant (MPE) and nonmalignant pleural effusion (non-MPE) in routine clinical practice. Objectives: To evaluate the utility of US features to predict malignancy in undiagnosed pleural effusions in a real-life clinical setting. Methods: The US features of 154 consecutive patients with a pleural effusion were prospectively assessed. Anonymous images were recorded by an operator blinded to the clinical and radiological results. The US findings were classified by independent reviewers and compared to the final diagnosis. Results: A total of 133 patients were included (age 67 ± 16 years; BMI 25.1 ± 4.6; 54.1% females). The final diagnosis was MPE in 66 cases and non-MPE in 67 cases. US had an overall sensitivity of 80.3%, a specificity of 83.6%, and positive and negative predictive values of 82.8 and 81.2%, respectively, for the detection of malignancy. US accuracy was 81.9%. The presence of pleural/diaphragmatic nodules, pleural/diaphragmatic thickness >10 mm, and a swirling sign was significantly different between both groups (p < 0.001). Lung air bronchogram sign and a septated US pattern were more common in non-MPE patients (p < 0.01). The existence of nodularity and the absence of air bronchograms were more likely to indicate malignancy (OR 29.0, 95% CI 7.65-110.08 and OR 10.4, 95% CI 1.65-65.752, respectively). Conclusions: In the presence of an undiagnosed pleural effusion, US morphological characteristics can aid in differentiating MPE from non-MPE. Pleural/diaphragmatic nodularity was the most relevant feature although no finding was pathognomonic of MPE.