New biomarkers are frequently being developed in laboratory settings for the early diagnosis of diseases. However, the assay can be so expensive to assess in some cases that the evaluation of a large number of assays becomes unfeasible. Under this setting pooling biospecimens becomes an appealing alternative. In this paper, we present the methodology to allow for general pooling strategies and different data structures, which include balanced and unbalanced pooling cases. An estimate of the area under the ROC curve of a single biomarker with its asymptotic mean and variance is provided. Furthermore, we develop a test statistic for comparing the areas under the ROC curves of two biomarkers. The methods are illustrated with data from a study evaluating biomarkers for coronary heart disease.