In the case of lung alveoli, the good contrast of the features allows the use of a fully automated image analysing procedure, yielding many of the desirable stereological parameters which can be estimated on optical microscopy slides: alveolar volume and surface densities, a form factor and average mean integral curvature. With our program, specially designed for the IBAS image analyser (Kontron Bildanalyse, Munich, B.R.D.), the whole automated process takes an average of 28 s per field. The image analyser-obtained stereological estimates compare well with those given by point and intersection counting methods. Some difficulties inherent in image analysers arise when measuring alveolar boundary lengths: a scheme is therefore proposed which eliminates the image-edge errors, with no sampling bias, by comparing three different frames inside each field. A solution to the problem of estimating automatically the positive and negative tangent counts, necessary for the curvature calculations, is also proposed. Our program has been applied routinely to the study of premature newborn rabbit lung alveoli, as part of a work dealing with an evaluation of the possibility of preventing the respiratory distress syndrome by the administration of heterologous surfactant.