We read with great interest the article by Gregor and colleagues on the management and survival of patients with lung cancer in Scotland diagnosed in 1995.1 The results were disappointing, but we congratulate them for their recognition of present conditions and for reporting the scientific analysis. In the 1990s several new chemotherapeutic drugs for lung cancer emerged, although the results of the large phase III studies were disappointing.2 ,3 It is fair to say that standard treatment for advanced lung cancer, especially for non-small cell lung cancer, is not yet established. Several well designed clinical trials have been reported in first class medical journals, but the prognosis of lung cancer is still poor. Published regimens for selected patients to define new study protocols may be inappropriate for use in clinical practice. Many of our patients are ordinary people who have several underlying illnesses and may be too sick to be enrolled ino clinical trials, and it is they who need treatment which can be applied in common practice. There is no disagreement on the point that the level of evidence obtained from the retrospective study of heterogeneous patients is low; however, we believe that a study with well analysed data of patients who are otherwise not eligible for randomised control trials also has clinical significance and would benefit such patients. We hope that the first class medical journals such asThorax continue to encourage, not only randomised control trials, but also case reports or retrospective studies to complement the area where strong evidence is unobtainable.