Optimal evaluation of residual masses of non Hodgkin's lymphomas (NHL) after chemotherapy is of major importance, and gallium scan (GS) is routinely used for this purpose, particularly for mediastinal sites. However, sensitivity and specificity of GS in this setting has been diversely appreciated and needs to be more accurately defined especially if radiotherapy is not planned. A retrospective analysis selected all the patients treated in a single institution for aggressive NHL who presented a residual mass in the mediastinum after chemotherapy and who were evaluated by GS. The value of GS for distinguishing true complete responses (CR) from partial responses (PR) was analyzed in patients who were either submitted to resection of their residual mass or followed up without further treatment after GS. A residual mass with mean perpendicular diameters measuring 4.1 cm x 2.8 cm was found in 42 patients and was GS positive in 8 cases and negative in 34 cases. After GS, radiotherapy was delivered to 10 patients, but 12 patients underwent resection of their residual mass and 20 were followed up without further treatment. In the patients who did not receive radiotherapy, 3 false positive and 6 false negative GS results were disclosed. The specificity and the sensitivity of GS were 88% and 25%, and its positive predictive value and negative predictive value 40% and 78%, respectively. GS was not sufficiently reliable to evaluate post chemotherapy residual masses. Surgical resection of residual masses should be considered particularly in young patients. Patients in true CR should be spared pointless radiotherapy and its late side effects, while patients in PR may benefit from further intensified chemotherapy followed by radiotherapy.