Multivariate models of survival have been established for both small cell and non-small cell lung cancers. So far, no study has focussed on squamous cell types. Previous demonstrations of the prognostic value of the tissue polypeptide antigen (TPA) and, partially, of the carcinoembryonic antigen (CEA) are based on univariate analyses of survival. These analyses do not account for the other prognostic factors. In the present study, we report the combined influence of various clinical and biological characteristics on the survival duration of 360 patients with a newly diagnosed squamous cell carcinoma of the lung. The study comprised 29 variables, including age, sex, smoking habit (SH), symptoms at diagnosis, the Karnofsky performance status (KPS), weight loss (WL), radiological findings, various disease extent parameters (DEP), CEA and TPA. Preliminary univariate analyses showed that 20 variables were survival-related. The Cox proportional hazards regression analysis selected stage of disease, KPS, TPA, WL, the existence of bone metastases, and SH as independent factors of prognosis (global chi-square: 122.40, P = 0.0000). A second multivariate analysis, performed with the same covariates but excluding DEP, revealed previous pulmonary diseases and CEA to be, in addition to KPS, TPA, SH, and WL the next most influential prognostic determinants. Also in squamous cell lung cancer, classifications based on the Cox's prediction equation may improve individual counseling and patient selection for therapeutic trials. In this malignancy, TPA shows an independent and strong prognostic significance while CEA shares informations of diverse other prognostic factors and seems to be less important.