Background and objective For previously treated recurrent non-small cell lung cancer, many studies have proven that inhibitors of the tyrosine kinase of epidermal growth factor receptor (EGFR-TKI), such as gefitinib and erlotinib can increase survival, especially in non-smoker adenocarcinoma. The objective of this study is to investigate the relationship between the characteristics of chest Computed tomography (CT) and objective response of gefitinib and erlotinib for recurrent lung adenocarcinoma. Methods Thirty-seven patients with recurrent advanced lung adenocarcinoma were treated from Jan 1, 2004 to Mar 31, 2007 in our department. The patients consisted of 17 males and 20 females previously treated with 1-6 cycle of platinum-based chemotherapy. The age changed from 54 to 85 yrs (media 67.5 yrs). The gefitinib and erlotinib was given 250 mg and 150 mg respectively per day until disease progression. The characteristics of chest CT were evaluated before therapy and reviewed every 2 month as follow up. The objective response was analyzed according to RECIST system. Results In 37 patients, 21 patients benefited (including 7 partial response and 14 stable disease) from the treatment. Chest CT characteristic analysis showed that, the benefit rate was 79.9% (15/19) in patients with greater than 3 metastasis of lung or diffused disease, vs. 33.3% (6/18) in others，P<0.05; the benefit rate was 73.1% (19/26) and 67.4% (31/46) respectively in patients with pleural effusion and moderate contrast disease focus in enhancement CT scanning, vs. 22.2% (2/9) and 37.5% (9/24) in patients without pleural effusion and non-enhancement respectively, P<0.05. But Cox regression analyses show no relationship between CT presentation and progress free survival (PFS) or overall survival (OS), HR, 0.766, P>0.05. Conclution For advanced lung adenocarcinoma, CT presentation of multiple metastasis of lung or diffused disease, pleural effusion and enhancing contrast disease focus may be a predictor for response when treated by EGFR-TKI. But those CT character does not represent as independent factor for prognosis.