In 407 patients with small-cell lung cancer (SCLC), 61 pretreatment variables were evaluated in a Cox multiple regression analysis to assess their prognostic value. All patients received short-term intensive regimens (cyclophosphamide, etoposide and methotrexate or ifosfamide and etoposide, both followed by thoracic irradiation if complete response was noted). Lactate dehydrogenase (p = 0.001), tumour stage (p = 0.0001), serum sodium (p = 0.0009), pretreatment Karnofsky performance score (p = 0.0121), alkaline phosphatase (p = 0.0186) and serum bicarbonate (p = 0.0321) were the important prognostic factors. Once these variables were taken into account no other variable provided additional prognostic information. A simple scoring system ("Manchester Score") using these variables was established and shows little loss of information compared to the Cox analysis. The score distinguishes 3 prognostic groups, the best of which contains all long-term survivors, whereas the bad prognostic group contains no patient surviving longer than one year. The scoring system may help to design new treatment strategies and may also facilitate the comparison of different studies.